analysis/05_figures_paper.R
In SC-COSMO/sccosmomcma: Stanford-CIDE COronavirus Simulation MOdel (SC-COSMO) for Mexico City Metropolitan Area Analysis
# 05_figures_paper.R
#-------------------------------------------------------------------------#
# This script creates the figures presented in the MCMA paper. #
# #
# Authors: #
# - Valeria Gracia Olvera, MsC #
#-------------------------------------------------------------------------#
rm(list=ls())
# Load packages and set constants -----------------------------------------
# Libraries
library(tidyverse)
library(ggplot2)
library(dplyr)
library(dampack)
library(ggpubr)
library(grid)
# Dates
n_date_projs <- "2020-12-13"
n_date_obs <- "2020-12-13"
n_time_stamp_hosp <- "2021-03-07"
# Load data ---------------------------------------------------------------
# Projections
load(paste0("output/05_projections_probabilistic_all_SQ_MCMA_",n_date_projs,".RData"))
load(paste0("output/05_projections_probabilistic_SQ_MCMA_",n_date_projs,".RData"))
df_out_total_prob_all <- df_out_mex_total_prob_all
df_out_total_prob <- df_out_mex_total_prob
# Observed data
load("data/l_targets_2020-12-13.RData")
# Hospitalization data
load(paste0("data/df_hosp_MCMA_2020-12-21.RData"))
# Plot specifications ----------------------------------------------------
# Names of outcomes
v_outcomes <- c(Hosp_tot = "Total hospitalizations",
CDR_prop = "CDR proportion",
Dcov_tot = "COVID deaths",
DXCumtot = "Cumulative cases",
InfCumtot = "Cumulative infections",
InfCumprop = "Cumulative infections proportion",
DXtot = "Detected cases",
DXDcov_tot = "Cumulative deaths",
NonICU_tot = "Hospitalizations without ventilator",
ICU_tot = "Hospitalizations with ventilator",
DXDIncCov_tot = "Incident deaths",
DXInctot = "Incident cases",
InfInctot = "Incident infections",
Inftot = "Prevalent infections",
Rt = "R effective",
Rec_tot = "Recovered prevalence",
Rec_prop = "Recovered prevalence proportion")
# State
state_i <- unique(df_out_total_prob$state)
# Abbreviation for state
abbrev_state <- unique(df_out_total_prob$s_code)
# Data time stamp
n_time_stamp <- as.Date(unique(df_out_total_prob$time_stamp))
# Last date
n_date_last <- as.Date("2020-12-07")
# Max date
max_date <- as.Date("2021-03-07")
# NPI date
date_NPI <- as.Date("2020-03-23")
# NPI update
date_NPI_update <- as.Date("2021-01-10")
# Holidays dates
n_holidays_start <- as.Date("2020-12-24")
# Text size
text_size <- 24
caption_size <- 20
annotation_size <- 17
bracket_size <- 1.1
label_bracket_size <- 6.8
arrow_size <- 3
# X axis formating
angle_x_axis <- 90
label_x_axis <- ""
# Date breaks
v_dates_breaks <- c(as.Date("2020-02-15"), as.Date("2020-03-01"),
as.Date("2020-03-15"), as.Date("2020-04-01"),
as.Date("2020-04-15"), as.Date("2020-05-01"),
as.Date("2020-05-15"), as.Date("2020-06-01"),
as.Date("2020-06-15"), as.Date("2020-07-01"),
as.Date("2020-07-15"), as.Date("2020-08-01"),
as.Date("2020-08-15"), as.Date("2020-09-01"),
as.Date("2020-09-15"), as.Date("2020-10-01"),
as.Date("2020-10-15"), as.Date("2020-11-01"),
as.Date("2020-11-15"), as.Date("2020-12-01"),
as.Date("2020-12-15"), as.Date("2021-01-01"),
as.Date("2021-01-15"), as.Date("2021-02-01"),
as.Date("2021-02-15"), as.Date("2021-03-01"))
# Set colors
jet.colors <- colorRampPalette(c("black", "#00007F", "blue", "#007FFF",
"cyan", "#7FFF7F", "yellow", "#FF7F00",
"red", "#7F0000"))
color_map <- jet.colors(100)
# Function
# Number of ticks for plots: dampack package
number_ticks <- function(n) {
function(limits) {
pretty(limits, n + 1)
}
}
# Wrangle data ------------------------------------------------------------
# Set vector of interventions and labels
v_interv_names <- c(BaseCase = "Policy A. Physical distancing: status quo; Schooling: not in-person",
IncreaseSD = "Policy B. Physical distancing: +27% compared to status quo; Schooling: not in-person",
IncreaseSDSchoolSD = "Policy C. Physical distancing: +27% compared to status quo; Schooling: in-person",
SchoolSD = "Policy D. Physical distancing: status quo; Schooling: in-person")
# Rename interventions
df_out_total_prob$Intervention <- as.character(df_out_total_prob$Intervention)
df_out_total_prob_all$Intervention <- as.character(df_out_total_prob_all$Intervention)
for(n_interv in names(v_interv_names)){
interv_type <- v_interv_names[which(names(v_interv_names) == n_interv)]
df_out_total_prob$Intervention[df_out_total_prob$intervention_type == n_interv] <- interv_type
df_out_total_prob_all$Intervention[df_out_total_prob_all$intervention_type == n_interv] <- interv_type
}
df_out_total_prob$Intervention <- factor(df_out_total_prob$Intervention,
levels = v_interv_names,
ordered = TRUE)
df_out_total_prob_all$Intervention <- factor(df_out_total_prob_all$Intervention,
levels = v_interv_names,
ordered = TRUE)
# Projection type
v_proj_labels <- c(SQ = "Base-case",
SA = "Lower Covid-19 risk in children")
# Projection type names
df_out_total_prob$proj_type_label <- ""
df_out_total_prob_all$proj_type_label <- ""
for(n_proj_type in c("SQ", "SA")){
df_out_total_prob$proj_type_label[df_out_total_prob$proj_type == n_proj_type] <- v_proj_labels[n_proj_type]
df_out_total_prob_all$proj_type_label[df_out_total_prob_all$proj_type == n_proj_type] <- v_proj_labels[n_proj_type]
}
df_out_total_prob$proj_type_label <- factor(df_out_total_prob$proj_type_label,
levels = v_proj_labels,
ordered = TRUE)
df_out_total_prob_all$proj_type_label <- factor(df_out_total_prob_all$proj_type_label,
levels = v_proj_labels,
ordered = TRUE)
# Create variable with base case labels
v_basecase_names <- c(StatusQuo = "End-of-year Holidays: status quo",
Holidays = "End-of-year Holidays: higher social contacts")
df_out_total_prob$BaseCase <- ""
df_out_total_prob_all$BaseCase <- ""
for(n_basecase in names(v_basecase_names)){
basecase_type <- v_basecase_names[which(names(v_basecase_names) == n_basecase)]
df_out_total_prob$BaseCase[df_out_total_prob$BaseCase_type == n_basecase] <- basecase_type
df_out_total_prob_all$BaseCase[df_out_total_prob_all$BaseCase_type == n_basecase] <- basecase_type
}
df_out_total_prob$BaseCase <- factor(df_out_total_prob$BaseCase,
levels = v_basecase_names,
ordered = TRUE)
df_out_total_prob_all$BaseCase <- factor(df_out_total_prob_all$BaseCase,
levels = v_basecase_names,
ordered = TRUE)
# Rename observed outcomes
df_out_total_prob$Outcome <- as.character(df_out_total_prob$Outcome)
df_out_total_prob$Outcome[df_out_total_prob$Outcome == "Cumulative detected cases"] <- "Cumulative cases"
df_out_total_prob$Outcome[df_out_total_prob$Outcome == "Incident detected cases"] <- "Incident cases"
df_out_total_prob$Outcome[df_out_total_prob$Outcome == "Detected COVID deaths"] <- "Cumulative deaths"
df_out_total_prob$Outcome[df_out_total_prob$Outcome == "Incident COVID19 deaths infections"] <- "Incident deaths"
df_out_total_prob_all$Outcome <- as.character(df_out_total_prob_all$Outcome)
df_out_total_prob_all$Outcome[df_out_total_prob_all$Outcome == "Cumulative detected cases"] <- "Cumulative cases"
df_out_total_prob_all$Outcome[df_out_total_prob_all$Outcome == "Incident detected cases"] <- "Incident cases"
df_out_total_prob_all$Outcome[df_out_total_prob_all$Outcome == "Detected COVID deaths"] <- "Cumulative deaths"
df_out_total_prob_all$Outcome[df_out_total_prob_all$Outcome == "Incident COVID19 deaths infections"] <- "Incident deaths"
## Offset hospitalizations ------------------------------------------------
offset_flag <- TRUE
# Modify data applying an offset
n_date_offset <- "2020-12-07"
# Outcomes of interest
v_hosp_outcomes <- c("Total hospitalizations",
"Hospitalizations without ventilator",
"Hospitalizations with ventilator")
if(offset_flag){
for(n_proj_type in c("SQ")){
for(n_hosp_outcome in c(v_hosp_outcomes)){
denominator <- df_hosp_MCMA$hosp_tot[df_hosp_MCMA$dates == n_date_offset]
offset_hosp_ratio <- mean(df_out_total_prob_all$value[df_out_total_prob_all$dates == n_date_offset &
df_out_total_prob_all$BaseCase_type == "StatusQuo" &
df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$intervention_type == "BaseCase" &
df_out_total_prob_all$Outcome == n_hosp_outcome])/denominator
# Modify data
df_out_total_prob_all$value[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome] <- df_out_total_prob_all$value[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome]/offset_hosp_ratio
df_out_total_prob_all$lb[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome] <- df_out_total_prob_all$lb[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome]/offset_hosp_ratio
df_out_total_prob_all$ub[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome] <- df_out_total_prob_all$ub[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome]/offset_hosp_ratio
}
}
}
# Covid-19 cases and deaths observed data
df_covid_obs_data <- rbind(l_targets$cases,
l_targets$cases_inc,
l_targets$deaths,
l_targets$deaths_inc) %>%
ungroup() %>%
filter(state %in% unique(df_out_total_prob$state)) %>%
rename(Outcome = Target,
s_code = abbrev_state,
dates = Date,
sd = se) %>%
select(-country, -county, -var_outcome, -series, -population, -Date0) %>%
mutate(type = "Observed",
BaseCase_type = "Observed",
Intervention = "Observed",
intervention_type = "Observed",
median = NA)
df_covid_obs_data <- df_covid_obs_data[,c("state", "s_code", "type", "BaseCase_type", "Intervention",
"intervention_type", "Outcome", "dates", "value", "median",
"sd", "lb", "ub","time_stamp")]
df_covid_obs_data$Outcome <- as.character(df_covid_obs_data$Outcome)
df_covid_obs_data$Outcome[df_covid_obs_data$Outcome == "Incident confirmed infections"] <- "Incident cases"
df_covid_obs_data$Outcome[df_covid_obs_data$Outcome == "Cumulative confirmed infections"] <- "Cumulative cases"
df_covid_obs_data$Outcome[df_covid_obs_data$Outcome == "Cumulative COVID19 deaths infections"] <-"Cumulative deaths"
df_covid_obs_data$Outcome[df_covid_obs_data$Outcome == "Incident COVID19 deaths infections"] <- "Incident deaths"
# Weekly data
df_covid_obs_data_week <- rbind(l_targets$cases,
l_targets$cases_inc,
l_targets$deaths,
l_targets$deaths_inc) %>%
ungroup() %>%
filter(state %in% unique(df_out_total_prob$state) &
Date <= n_date_last) %>%
rename(Outcome = Target,
s_code = abbrev_state,
dates = Date,
sd = se) %>%
select(-country, -county, -var_outcome, -series, -population, -Date0) %>%
mutate(type = "Observed",
BaseCase_type = "Observed",
Intervention = "Observed",
intervention_type = "Observed",
median = NA) %>%
group_by(state, type, Outcome, DateLast, time_stamp, s_code,
BaseCase_type, Intervention, intervention_type,
week = cut.Date(dates, breaks = "week")) %>%
summarise(value_week = mean(value),
lb_week = mean(lb),
ub_week = mean(ub),
sd_week = mean(sd)) %>%
rename(value = value_week,
lb = lb_week,
ub = ub_week,
sd = sd_week,
dates = week) %>%
mutate(median = NA) %>%
ungroup()
df_covid_obs_data_week <- df_covid_obs_data_week[,c("state", "s_code", "type", "BaseCase_type", "Intervention",
"intervention_type", "Outcome", "dates", "value", "median",
"sd", "lb", "ub","time_stamp")]
df_covid_obs_data_week$Outcome <- as.character(df_covid_obs_data_week$Outcome)
df_covid_obs_data_week$Outcome[df_covid_obs_data_week$Outcome == "Incident confirmed infections"] <- "Incident cases"
df_covid_obs_data_week$Outcome[df_covid_obs_data_week$Outcome == "Cumulative confirmed infections"] <- "Cumulative cases"
df_covid_obs_data_week$Outcome[df_covid_obs_data_week$Outcome == "Cumulative COVID19 deaths infections"] <-"Cumulative deaths"
df_covid_obs_data_week$Outcome[df_covid_obs_data_week$Outcome == "Incident COVID19 deaths infections"] <- "Incident deaths"
# ABSTRACT ----------------------------------------------------------------
# # Interventions
# select_intervention <- c("SchoolSD",
# "IncreaseSD",
# "IncreaseSDSchoolSD",
# "BaseCase")
#
# df_AbstractCases <- df_out_total_prob_all %>%
# filter(dates >= "2020-12-07" &
# # BaseCase_type=="Holidays" &
# intervention_type %in% select_intervention &
# Outcome == "Incident cases") %>%
# group_by(county, type, BaseCase_type, Intervention, intervention_type, Outcome,
# simulation, proj_type, proj_type_label) %>%
# summarise(cum_value = sum(value)) %>%
# group_by(county, type, BaseCase_type, Intervention, intervention_type, Outcome,
# proj_type, proj_type_label) %>%
# summarise(value = round(mean(cum_value/1e6),2),
# sd = round(sd(cum_value/1e6),2),
# lb = round(quantile(cum_value/1e6, probs = 0.025, names = FALSE), 2),
# ub = round(quantile(cum_value/1e6, probs = 0.975, names = FALSE), 2)) %>%
# mutate(time_stamp = n_time_stamp) %>%
# rename(state = county)
#
# df_AbstractHosp <- df_out_total_prob_all %>%
# filter(dates >= "2020-12-07" &
# intervention_type %in% select_intervention &
# Outcome == "Total hospitalizations") %>%
# group_by(county, type, BaseCase_type, Intervention, intervention_type, Outcome,
# proj_type, proj_type_label, dates) %>%
# summarise(mean = round(mean(value),2),
# sd = round(sd(value),2),
# lb = round(quantile(value, probs = 0.025, names = FALSE), 2),
# ub = round(quantile(value, probs = 0.975, names = FALSE), 2)) %>%
# mutate(time_stamp = n_time_stamp) %>%
# rename(state = county)
#
# # First value:
# df_AbstractHosp[df_AbstractHosp$BaseCase_type == "Holidays" &
# df_AbstractHosp$intervention_type == "BaseCase" &
# df_AbstractHosp$mean == max(df_AbstractHosp$mean[df_AbstractHosp$BaseCase_type == "Holidays" &
# df_AbstractHosp$intervention_type == "BaseCase"]),]
# # Second value
# df_AbstractHosp[df_AbstractHosp$BaseCase_type == "StatusQuo" &
# df_AbstractHosp$intervention_type == "IncreaseSDSchoolSD" &
# df_AbstractHosp$mean == max(df_AbstractHosp$mean[df_AbstractHosp$BaseCase_type == "StatusQuo" &
# df_AbstractHosp$intervention_type == "IncreaseSDSchoolSD"]),]
# FIGURE 1: cases and deaths ----------------------------------------------
# Vector of outcomes of interest
v_outcome <- c("Incident cases",
"Incident deaths",
"Cumulative cases",
"Cumulative deaths")
# Wrangle projections
df_outcome_int <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
Outcome %in% v_outcome &
intervention_type == "BaseCase" &
dates <= max_date) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, proj_type, proj_type_label) %>%
summarise(mean = mean(value),
median = quantile(value, probs = 0.5, names = FALSE),
sd = sd(value),
lb = quantile(value, probs = 0.025, names = FALSE),
ub = quantile(value, probs = 0.975, names = FALSE)) %>%
rename(value = mean,
state = county) %>%
mutate(time_stamp = n_time_stamp) %>%
select(state, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, value, lb, ub, time_stamp,
proj_type, proj_type_label) %>%
ungroup()
df_outcome_int$Intervention <- "Model-predicted"
# Observed data
df_outcome_obs <- rbind.data.frame(
subset(df_covid_obs_data_week, Intervention == "Observed") %>%
mutate(BaseCase_type = "Holidays",
BaseCase = v_basecase_names["Holidays"],
proj_type = "SQ",
proj_type_label = "Base-case"),
subset(df_covid_obs_data_week, Intervention == "Observed") %>%
mutate(BaseCase_type = "StatusQuo",
BaseCase = v_basecase_names["StatusQuo"],
proj_type = "SQ",
proj_type_label = "Base-case")) %>%
select(state, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, value, lb, ub, time_stamp,
proj_type, proj_type_label)
df_outcome_obs$Intervention <- as.factor(df_outcome_obs$Intervention)
# Create and save data.frame
df_fig1_CasesDeaths <- rbind.data.frame(df_outcome_int,
df_outcome_obs)
# Save data
save(df_fig1_CasesDeaths,
file = paste0("figs/figs_paper/data_frames/df_fig1_CasesDeaths.RData"))
## Plot specifications ----------------------------------------------------
# Colors
v_names_colours_obs <- as.character(levels(df_fig1_CasesDeaths$Intervention))
v_colors_colour <- rev(c("#a13023","#659583"))
names(v_colors_colour) <- v_names_colours_obs
# Vector of colors to fill
v_colors_fill <- rev(c(NA,"#659583"))
names(v_colors_fill) <- v_names_colours_obs
# Shape
v_shape <- rev(c(16,NA))
names(v_shape) <- v_names_colours_obs
# Line
v_linetype <- rev(c("blank","solid"))
names(v_linetype) <- v_names_colours_obs
# For annotation_custom
npi_date <- grobTree(textGrob("Date of NPI implementation",
x = 0.12,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
calib_day <- grobTree(textGrob("Last day used for calibration",
x = 0.756,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
# Save plots in list
l_plots_fig1 <- vector(mode = "list", length = 4)
i <- 1
for(n_proj_type in c("SQ")){ # n_proj_type = "SA"
for(outcome in v_outcome){
# Abbreviation for outcome to save the plot
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Filter data to outcome and type of data
df_obs_temp <- subset(df_fig1_CasesDeaths,
Outcome == outcome &
Intervention == "Observed" &
proj_type == n_proj_type)
df_int_temp <- subset(df_fig1_CasesDeaths,
Outcome == outcome &
Intervention == "Model-predicted" &
proj_type == n_proj_type)
# Bracket position
y_end <- max(df_obs_temp$ub,df_int_temp$ub)
# Plot
l_plots_fig1[[i]] <- ggplot(data = df_obs_temp) +
geom_point(mapping = aes(x = dates,
y = value,
colour = Intervention,
fill = Intervention),
shape = 16,
size = 3,
alpha = 0.6) +
geom_line(data = df_int_temp,
aes(x = dates,
y = value,
colour = Intervention),
size = 1.1) + #0.9) +
geom_ribbon(data = df_int_temp,
aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = Intervention),
alpha = 0.4,
colour = NA) +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour,2)) +
scale_fill_manual(values = rep(v_colors_fill, 2)) +
guides(fill = guide_legend(title = ""),
color = guide_legend(title = "",
ncol = 2,
override.aes = list(shape = v_shape,
linetype = v_linetype,
color = v_colors_colour,
fill = v_colors_fill))) +
geom_vline(xintercept = c(as.numeric(as.Date("2020-03-23")),
as.numeric(n_date_last)),
show.legend = TRUE,
size = 0.5,
linetype = rep(c("dashed", "twodash"), 2),
color = "#393F3E") +
annotation_custom(npi_date) +
geom_bracket(data = NULL, xmin = n_date_last,
xmax = max_date,
y.position = y_end*1.04,
label = "Projections",
label.size = label_bracket_size,
size = bracket_size) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks
) +
scale_y_continuous(outcome,
breaks = number_ticks(12),
labels = scales::comma) + # trans = "log"
theme(text = element_text(size = text_size),
axis.text.x = element_text(angle = angle_x_axis,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "bottom",
legend.justification = "center",
legend.text = element_text(size = 22),
legend.direction = "horizontal",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent",
size = unit(3, "cm"))) +
coord_cartesian(ylim = c(0, y_end*1.1))
i <- i + 1
}
i <- 1
# Combine graphs
plot_inc <- ggarrange(l_plots_fig1[[i]],
NULL,
l_plots_fig1[[i+1]],
ncol = 1,
common.legend = T,
legend = "none",
heights = c(1,0.05, 1),
labels = c("A","","B"),
hjust = 0,vjust = 1.5,
font.label = list(size = 26, face = "bold"))
# Cumulative
plot_cum <- ggarrange(l_plots_fig1[[i+2]],
NULL,
l_plots_fig1[[i+3]],
ncol = 1,
common.legend = T,
legend = "bottom",
heights = c(1,0.05, 1),
labels = c("C","","D"),
hjust = 0,vjust = 1.5,
font.label = list(size = 26, face = "bold"))
plot_fig1 <- ggarrange(plot_inc,
plot_cum,
ncol = 1)
# Save plot
ggsave(paste0("figs/figs_paper/fig1_CasesDeaths_",n_proj_type,".jpg"),
width = 18, height = 32, dpi = 300)
}
# FIGURE 2: Hospitalizations ----------------------------------------------
# Vector of outcomes of interest
outcome <- "Total hospitalizations"
# Offset flag
offset_flag <- TRUE
# Hosp projected data
df_Hosp_proj <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
intervention_type == "BaseCase" &
Outcome == outcome &
dates <= max_date) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, proj_type, proj_type_label) %>%
summarise(mean = round(mean(value),0),
median = round(quantile(value, probs = 0.5, names = FALSE),0),
sd = round(sd(value),0),
lb = round(quantile(value, probs = 0.025, names = FALSE),0),
ub = round(quantile(value, probs = 0.975, names = FALSE),0)) %>%
rename(value = mean,
state = county) %>%
mutate(time_stamp = n_time_stamp,
cap_tot = NA) %>%
select(state, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, value, lb, ub, cap_tot, time_stamp,
proj_type, proj_type_label)
df_Hosp_proj$Intervention <- "Model-predicted"
# Observed data
df_hosp_tot <- df_hosp_MCMA %>%
filter(dates <= n_date_last)
df_hosp_tot$dates <- as.Date(df_hosp_tot$dates)
df_hosp_tot <- df_hosp_tot[order(df_hosp_tot$dates,decreasing = F),]
df_hosp_tot <- df_hosp_tot %>%
mutate(state = state_i,
type = "Observed",
Outcome = outcome,
Intervention = "Observed",
intervention_type = "Observed",
lb = NA,
ub = NA, time_stamp = as.Date(n_date_obs)
) %>%
filter(dates <= n_time_stamp_hosp) %>%
rename(value = hosp_tot)
df_hosp_tot$Intervention <- as.factor(df_hosp_tot$Intervention)
# Create data.frame with two set of base-case
df_hosp_observed <- rbind.data.frame(df_hosp_tot %>% mutate(BaseCase_type = "Holidays",
BaseCase = v_basecase_names["Holidays"],
proj_type = "SQ",
proj_type_label = "Base-case"),
df_hosp_tot %>% mutate(BaseCase_type = "StatusQuo",
BaseCase = v_basecase_names["StatusQuo"],
proj_type = "SQ",
proj_type_label = "Base-case")) %>%
select(state, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, value, lb, ub, cap_tot, time_stamp,
proj_type, proj_type_label
)
# Bind data.frames
df_fig2_HospTot <- rbind.data.frame(df_Hosp_proj,
df_hosp_observed)
# Save
if(offset_flag){
save(df_fig2_HospTot,
file = paste0("figs/figs_paper/data_frames/df_fig2_HospTot_offset.RData"))
}else{
save(df_fig2_HospTot,
file = paste0("figs/figs_paper/data_frames/df_fig2_HospTot.RData"))
}
## Plot specifications ----------------------------------------------------
# Order interventions
v_interv <- as.character(unique(df_fig2_HospTot$Intervention))
df_fig2_HospTot$Intervention <- ordered(df_fig2_HospTot$Intervention,
rev(v_interv))
# Colors
v_names_colours_obs <- as.character(levels(df_fig2_HospTot$Intervention))
v_colors_colour <- c("#a13023", "#659583")
names(v_colors_colour) <- v_names_colours_obs
# Vector of colors to fill
v_colors_fill <- c("#a13023","#659583")
names(v_colors_fill) <- v_names_colours_obs
# Line
v_linetype <- c("solid","solid")
names(v_linetype) <- v_names_colours_obs
# For annotation_custom
npi_date <- grobTree(textGrob("Date of NPI implementation",
x = 0.12,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
calib_day <- grobTree(textGrob("Last day used for calibration",
x = 0.675,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
# List of plots
l_plot_fig2 <- vector(mode = "list", length = 2)
i <- 1
for(n_proj_type in c("SQ")){
# Abbreviation for outcome to save the plot
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Filter outcome of interest
df_obs_temp <- subset(df_fig2_HospTot,
Intervention == "Observed" &
proj_type == n_proj_type)
df_int_temp <- subset(df_fig2_HospTot,
Intervention == "Model-predicted" &
proj_type == n_proj_type)
# Bracket position
y_end <- max(df_obs_temp$value,df_int_temp$ub)
# Plot
l_plot_fig2[[i]] <- ggplot(data = df_obs_temp) +
geom_area(position = "stack",
aes(x = dates,
y = value,
colour = Intervention,
fill = Intervention),
alpha = 0.4) +
geom_line(data = df_int_temp,#subset(df_outcome_int, dates >= max(df_hosp_tot$dates)),
aes(x = dates,
y = value,
colour = Intervention),
linetype = "solid") +
geom_ribbon(data = df_int_temp, #subset(df_outcome_int, dates >= max(df_hosp_tot$dates)),
aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = Intervention),
alpha = 0.4,
colour = NA) +
geom_line(data = df_obs_temp,
aes(x = dates,
y = cap_tot),
size = 1.2,
color = "#393F3E") +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour, 2)) +
scale_fill_manual(values = rep(v_colors_fill, 2)) +
# scale_color_manual(values = v_colors_colour) +
# scale_fill_manual(values = v_colors_fill) +
guides(fill = guide_legend(title = "",
reverse = T),
color = guide_legend(title = "",
ncol = 2,
override.aes = list(color = v_colors_colour,
fill = v_colors_fill),
reverse = T)) +
geom_vline(xintercept = c(as.numeric(as.Date("2020-03-23")), as.numeric(n_date_last)),
show.legend = TRUE,
size = 0.5,
linetype = rep(c("dashed", "twodash"), 2),
# linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(npi_date) +
# annotation_custom(calib_day) +
annotate("text",
label = paste0("Current capacity: ",
last(df_obs_temp$cap_tot)," beds"),
x = as.Date("2020-09-13"),
y = last(df_obs_temp$cap_tot) + 2000,
size = 6,
colour = "#393F3E") +
geom_bracket(data = NULL, xmin = n_date_last,
xmax = max_date,
y.position = y_end*1.04,
label = "Projections",
label.size = label_bracket_size,
size = bracket_size) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks,
) +
scale_y_continuous(outcome,
breaks = number_ticks(8),
labels = scales::comma) + # trans = "log"
# labs(title = "COVID-19 hospitalizations with ventilator in MCMA, Mexico" # , caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = caption_size),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "bottom",
legend.justification = "center",
legend.direction = "vertical",
legend.text = element_text(size = 22),
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent",
size = unit(3, "cm"))) +
coord_cartesian(ylim = c(0, y_end*1.1))
# Save plot
if(offset_flag){
ggsave(paste0("figs/figs_paper/fig2_HospTot_offset_",n_proj_type,".jpg"),
width = 16, height = 9, dpi = 300)
}else{
ggsave(paste0("figs/figs_paper/fig2_HospTot_",n_proj_type,".jpg"),
width = 16, height = 9, dpi = 300)
}
i <- i + 1
}
# FIGURE 3: interventions cases and deaths --------------------------------
# Outcomes
v_outcome <- c("Incident deaths",
"Incident cases")
# Interventions
select_intervention <- c("SchoolSD",
"IncreaseSD",
"IncreaseSDSchoolSD",
"BaseCase")
# Data.frame
df_fig3_CasesDeaths_interv <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
Outcome %in% v_outcome &
intervention_type %in% select_intervention &
dates > n_date_last &
dates <= max_date) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, proj_type, proj_type_label) %>%
summarise(mean = mean(value),
median = quantile(value, probs = 0.5, names = FALSE),
sd = sd(value),
lb = quantile(value, probs = 0.025, names = FALSE),
ub = quantile(value, probs = 0.975, names = FALSE)) %>%
rename(value = mean)
# Order outcomes
v_outcom <- as.character(unique(df_fig3_CasesDeaths_interv$Outcome))
df_fig3_CasesDeaths_interv$Outcome <- ordered(df_fig3_CasesDeaths_interv$Outcome,
v_outcom[c(2,1)])
# Save data.frame
save(df_fig3_CasesDeaths_interv,
file = paste0("figs/figs_paper/data_frames/df_fig3_CasesDeaths_interv.RData"))
## Plot specifications ----------------------------------------------------
# Colors
v_names_colours <- c(as.character(levels(df_fig3_CasesDeaths_interv$Intervention)))
v_colors_colour <- c("#0F57CA", "#008450", "#FF9900", "#B81D13")
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- v_colors_colour
names(v_colors_fill) <- v_names_colours
# Vector of colors to shape
v_colors_shape <- c(15, 16, 17, 18)
names(v_colors_shape) <- v_names_colours
# Bracket position
y_end <- max(df_fig3_CasesDeaths_interv$value)
NPI_update <- grobTree(textGrob("Policy start date",
x = 0.14,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
# List of plots
i <- 1
l_plots_fig3 <- vector(mode = "list", length = 4)
for(n_proj_type in c("SQ")){
# Plot incident cases
l_plots_fig3[[i]] <- ggplot(
data = subset(df_fig3_CasesDeaths_interv, Outcome == "Incident cases" &
proj_type == n_proj_type),
aes(x = dates,
y = value,
colour = Intervention,
shape = Intervention)) +
# geom_bar(stat = "identity", position = position_dodge(width = 5),
# width = 3) +
geom_line(size = 0.9) +
# geom_point(size = 3) +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour, 2)) +
scale_shape_manual(values = rep(v_colors_shape, 2)) +
geom_vline(xintercept = c(as.numeric(date_NPI_update)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed"),
# linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(NPI_update) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = "week") +
scale_y_continuous("Cases",#"Difference to case base scenario",
breaks = number_ticks(8),
# labels = scales::percent
labels = scales::comma) + # trans = "log"
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = caption_size+2),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
legend.position = "bottom", # c(0.44,0.5),
legend.text = element_text(size = 20),
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
# Plot incident deaths
l_plots_fig3[[i+1]] <- ggplot(
data = subset(df_fig3_CasesDeaths_interv, Outcome == "Incident deaths" &
proj_type == n_proj_type),
aes(x = dates,
y = value,
colour = Intervention,
shape = Intervention)) +
geom_line(size = 0.9) +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour, 2)) +
scale_shape_manual(values = rep(v_colors_shape, 2)) +
geom_vline(xintercept = c(as.numeric(date_NPI_update)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed"),
# linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(NPI_update) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = "week") +
scale_y_continuous("Deaths",#"Difference to case base scenario",
breaks = number_ticks(8),
# labels = scales::percent
labels = scales::comma) + # trans = "log"
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = caption_size+2),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
legend.position = "bottom", # c(0.44,0.5),
legend.text = element_text(size = 20),
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
## Combine graphs ---------------------------------------------------------
plot_fig3 <- ggarrange(l_plots_fig3[[i]],
NULL,
l_plots_fig3[[i+1]],
ncol = 1,
common.legend = T,
legend = "bottom", heights = c(1, 0.05, 1),
labels = c("A","","B"), hjust = 0,vjust = 1.5,
font.label = list(size = 26, face = "bold"))
# Save plot
ggsave(paste0("figs/figs_paper/fig3_CasesDeaths_interv_",n_proj_type,".jpg"),
width = 16, height = 18, dpi = 300)
i <- i + 2
}
# FIGURE 4: interventions total hospitalizations --------------------------
# Interventions
select_intervention <- c("SchoolSD",
"IncreaseSD",
"IncreaseSDSchoolSD",
"BaseCase")
# Offset flag
offset_flag <- TRUE
# Data.frame
df_hosp_all_INT_fig4 <- df_out_total_prob_all %>%
filter(Outcome == "Total hospitalizations" &
intervention_type %in% select_intervention)
df_hosp_all_INT_fig4 <- df_hosp_all_INT_fig4[order(df_hosp_all_INT_fig4$dates, decreasing = F),]
# Observed data
df_hosp_fig4 <- df_hosp_MCMA %>%
filter(dates <= n_date_last)
df_hosp_fig4$dates <- as.Date(df_hosp_fig4$dates)
df_hosp_fig4 <- df_hosp_fig4[order(df_hosp_fig4$dates,decreasing = F),]
df_hosp_fig4 <- df_hosp_fig4 %>%
mutate(Intervention = "Observed") %>%
filter(dates <= n_date_last)
# Data.frame
df_fig4_HospTot_interv <- df_hosp_all_INT_fig4 %>%
filter(county == state_i &
intervention_type %in% select_intervention) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, proj_type, proj_type_label) %>%
summarise(mean = round(mean(value),0),
median = round(quantile(value, probs = 0.5, names = FALSE),0),
sd = sd(value),
lb = round(quantile(value, probs = 0.025, names = FALSE),0),
ub = round(quantile(value, probs = 0.975, names = FALSE),0)) %>%
rename(value = mean,
state = county
) %>%
filter(dates >= n_date_last) %>%
mutate(cap_tot = unique(df_hosp_observed$cap_tot[df_hosp_observed$dates==n_date_last]),
time_stamp = n_time_stamp)
df_fig4_HospTot_interv$dates <- as.Date(df_fig4_HospTot_interv$dates)
# Save
if(offset_flag){
save(df_fig4_HospTot_interv,
file = paste0("figs/figs_paper/data_frames/df_fig4_HospTot_interv_offset.RData"))
}else{
save(df_fig4_HospTot_interv,
file = paste0("figs/figs_paper/data_frames/df_fig4_HospTot_interv.RData"))
}
## Plot specifications ----------------------------------------------------
# Colors
v_names_colours <- c(as.character(levels(df_fig4_HospTot_interv$Intervention)))
v_colors <- c("#0F57CA", "#008450", "#FF9900", "#B81D13")
# Vector of colors to colour
v_colors_colour <- v_colors
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- v_colors
names(v_colors_fill) <- v_names_colours
# Vector of colors to shape
v_colors_shape <- c(15, 16, 17, 18)
names(v_colors_shape) <- v_names_colours
# Annotations
NPI_update <- grobTree(textGrob("Policy start date",
x = 0.14,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
# List of plots
l_plot_fig4 <- vector(mode = "list", length = 2)
i <- 1
for(n_proj_type in c("SQ")){
# Plot
l_plot_fig4[[i]] <- ggplot(
data = subset(df_fig4_HospTot_interv, proj_type == n_proj_type),
aes(x = dates,
y = value,
colour = Intervention)) +
# geom_bar(stat = "identity", position = position_dodge(width = 5),
# width = 3) +
geom_line(size = 0.9) +
# geom_point(size = 3) +
geom_line(aes(x = dates,
y = cap_tot),
size = 1.2,
color = "#393F3E" #, linetype = "twodash"
) +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour, 2)) +
# scale_shape_manual(values = rep(v_colors_shape, 2)) +
geom_vline(xintercept = c(as.numeric(date_NPI_update)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed"),
# linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(NPI_update) +
annotate("text",
label = paste0("Current capacity: ",
unique(df_hosp_observed$cap_tot[df_hosp_observed$dates==n_date_last])," beds"),
x = as.Date("2020-12-27"),
y = unique(df_hosp_observed$cap_tot[df_hosp_observed$dates==n_date_last]) + 1000,
size = 6,
colour = "#393F3E") +
# scale_fill_manual(values = rep(v_colors_fill, 2)) +
# scale_color_manual(values = v_colors_colour) +
# scale_fill_manual(values = v_colors_fill) +
# guides(fill = guide_legend(title = ""),
# color = guide_legend(title = "",
# ncol = 1,
# override.aes = list(color = v_colors_colour,
# fill = v_colors_fill))) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = "week") +
scale_y_continuous("Total hospitalizations",
breaks = number_ticks(8),
# labels = scales::percent
labels = scales::comma) + # trans = "log"
# labs(# title = "COVID-19 incident cases in MCMA, Mexico",
# caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = caption_size),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
legend.position = "bottom", # c(0.44,0.5),
legend.text = element_text(size = 20),
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
# Save plot
if(offset_flag){
ggsave(paste0("figs/figs_paper/fig4_HospTot_interv_offset_",n_proj_type,".jpg"),
width = 16, height = 9, dpi = 300)
}else{
ggsave(paste0("figs/figs_paper/fig4_HospTot_interv_",n_proj_type,".jpg"),
width = 16, height = 9, dpi = 300)
}
i <- i + 1
}
# FIGURE 5: exceed hosp capacity ------------------------------------------
# Interventions
select_intervention <- c("SchoolSD",
"IncreaseSD",
"IncreaseSDSchoolSD",
"BaseCase")
# Vector of outcomes of interest
outcome <- "Total hospitalizations"
# Hospital capacity
v_hosp_capacity <- c(Hosp_tot = 9667,
NonICU_tot = 7008,
ICU_tot = 2659)
# Interventions data
df_fig5_ExceedHospCap <- df_out_total_prob_all %>%
filter(Outcome == outcome &
intervention_type %in% select_intervention &
dates >= n_date_last) %>%
mutate(cap_tot = df_hosp_MCMA$cap_tot[df_hosp_MCMA$dates == n_date_last]) %>%
mutate(exceed_cap = ifelse(value > cap_tot, 1, 0)) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention, intervention_type,
Outcome, dates, proj_type, proj_type_label) %>%
mutate(exceed_cap_count = sum(exceed_cap),
exceed_cap_prop = exceed_cap_count*100/1000) %>%
summarise(value = mean(exceed_cap_prop),
median = quantile(exceed_cap_prop, probs = 0.5, names = FALSE),
sd = sd(exceed_cap_prop),
lb = quantile(exceed_cap_prop, probs = 0.025, names = FALSE),
ub = quantile(exceed_cap_prop, probs = 0.975, names = FALSE)) %>%
rename(state = county)
# Rename intervention_label
df_fig5_ExceedHospCap$intervention_label <- ""
for(n_interv in names(v_interv_labels)){
df_fig5_ExceedHospCap$intervention_label[df_fig5_ExceedHospCap$intervention_type == n_interv] <- v_interv_labels[n_interv]
}
df_fig5_ExceedHospCap$intervention_label <- factor(df_fig5_ExceedHospCap$intervention_label,
levels = v_interv_labels,
ordered = TRUE)
# Save
if(offset_flag){
save(df_fig5_ExceedHospCap,
file = paste0("figs/figs_paper/data_frames/df_fig5_ExceedHospCap_",abbrev_outcome,"_offset.RData"))
}else{
save(df_fig5_ExceedHospCap,
file = paste0("figs/figs_paper/data_frames/df_fig5_ExceedHospCap_",abbrev_outcome,".RData"))
}
# Plot specifications -----------------------------------------------------
# Caption
# Set vector of interventions and labels
v_interv_labels <- c(BaseCase = "Policy A",
IncreaseSD = "Policy B",
IncreaseSDSchoolSD = "Policy C",
SchoolSD = "Policy D"
)
plot_caption <- ""
for(intv in unique(df_fig5_ExceedHospCap$Intervention)){
plot_caption <- paste0(plot_caption, paste0(intv,"\n"))
}
## Plot -------------------------------------------------------------------
# List of plots
l_plots_fig5 <- vector(mode = "list", length = 2)
i <- 1
for(n_proj_type in c("SQ")){
# Abbrev outcome
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Plot
l_plots_fig5[[i]] <- ggplot(data = subset(df_fig5_ExceedHospCap, proj_type == n_proj_type),
aes(x = dates,
y = intervention_label,
fill = value)) +
geom_raster() +
labs(y = "", fill = "Probability of\nexceeding hospital\ncapacity (%)") +
scale_fill_gradientn(colours = color_map,
breaks = c(0.0, 25.0, 50.0, 75.0, 100.0),
labels = c("0", "25", "50", "75", "100"),
limits = c(0,100)) + # Hawre's jet map
facet_wrap( ~ BaseCase, ncol = 1) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
# breaks = seq.Date(from = first(df_hosp_all_INT_daily$week),
# to = last(df_hosp_all_INT_daily$week),
# by = "week")
breaks = "week") +
labs(caption = plot_caption,
title = outcome) +
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = 19),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "right",
legend.title = element_text(size = 18),
legend.justification = "center",
legend.direction = "vertical",
legend.key.width = unit(0.9, "cm"),
legend.key.height = unit(1.2, "cm"),
legend.title.align = 0,
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
if(offset_flag){
ggsave(paste0("figs/figs_paper/fig5_ExceedHospCap_",abbrev_outcome,"_",n_proj_type,"_offset.jpg"),
width = 16, height = 11, dpi = 300)
}else{
ggsave(paste0("figs/figs_paper/fig5_ExceedHospCap_",abbrev_outcome,"_",n_proj_type,".jpg"),
width = 16, height = 11, dpi = 300)
}
i <- i + 1
}
# First value in abstract:
df_fig5_ExceedHospCap[df_fig5_ExceedHospCap$BaseCase_type == "Holidays" &
df_fig5_ExceedHospCap$intervention_type == "BaseCase" &
df_fig5_ExceedHospCap$dates == "2021-01-25",]
# Second value in abstract:
df_fig5_ExceedHospCap[df_fig5_ExceedHospCap$BaseCase_type == "StatusQuo" &
df_fig5_ExceedHospCap$intervention_type == "IncreaseSDSchoolSD" &
df_fig5_ExceedHospCap$dates == "2021-01-19",]
# Table S2: cases and deaths ----------------------------------------------
v_outcome <- c("Incident cases",
"Cumulative cases",
"Incident deaths",
"Cumulative deaths")
# Interventions
select_intervention <- c("SchoolSD",
"IncreaseSD",
"IncreaseSDSchoolSD",
"BaseCase")
# Intervention data
df_tableS2 <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
intervention_type %in% select_intervention &
Outcome %in% v_outcome &
dates == max_date) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, proj_type, proj_type_label) %>%
summarise(mean = round(mean(value),0),
median = quantile(value, probs = 0.5, names = FALSE),
sd = round(sd(value),0),
lb = round(quantile(value, probs = 0.025, names = FALSE),0),
ub = round(quantile(value, probs = 0.975, names = FALSE),0)) %>%
rename(value = mean,
state = county) %>%
mutate(time_stamp = n_time_stamp) %>%
select(state, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, value, lb, ub, time_stamp,
proj_type, proj_type_label)
# Save
save(df_tableS2, file = "figs/figs_paper/data_frames/df_tableS2.RData")
# Figure S4: Effective Reproduction Number --------------------------------
## Status quo -------------------------------------------------------------
# Outcome
outcome <- "R effective"
outcome_eng <- "R effective"
# Data.frame
df_outcome_Rt <- df_out_total_prob %>%
ungroup() %>%
filter(state == state_i &
Outcome == outcome_eng &
dates <= max_date)
# Create data.frame for BaseCase
df_Rt_BaseCase <- subset(df_outcome_Rt, intervention_type == "BaseCase")
df_Rt_BaseCase$Intervention <- "Model-projected"
# Create data.frame for Interventions
df_RTt_INT <- subset(df_outcome_Rt, intervention_type %in% c("SchoolSD", "IncreaseSD","IncreaseSDSchoolSD"))
# Save data.frame
df_figS4_Rt <- rbind.data.frame(df_Rt_BaseCase,
df_RTt_INT)
save(df_figS4_Rt,
file = paste0("figs/figs_paper/data_frames/df_figS4_Rt.RData"))
# List of plots
l_plots_S4 <- vector(mode = "list", length = 4)
i <- 1
for(n_proj_type in c("SQ")){
### Plot: SQ --------------------------------------------------------------
# Abbreviation for outcome to save the plot
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Name of colors
v_names_colours <- as.character(levels(df_Rt_BaseCase$BaseCase))
# Colors
v_colors_colour <- c("#659583", "#a13023" )
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- v_colors_colour
names(v_colors_fill) <- v_names_colours
# Line
v_linetype <- c("solid", "solid")
names(v_linetype) <- v_names_colours
# Bracket position
y_end <- max(df_Rt_BaseCase$ub)
# For annotation_custom
npi_date <- grobTree(textGrob("Date of NPI implementation",
x = 0.115,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
calib_day <- grobTree(textGrob("Last day used for calibration",
x = 0.675,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
# Plot
l_plots_S4[[i]] <- ggplot(data = subset(df_Rt_BaseCase, proj_type == n_proj_type)) +
geom_line(aes(x = dates,
y = value,
colour = BaseCase),
size = 0.9) +
geom_ribbon(aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = BaseCase),
alpha = 0.4,
colour = NA) +
scale_color_manual(values = v_colors_colour) +
scale_fill_manual(values = v_colors_fill) +
guides(fill = guide_legend(title = ""),
color = guide_legend(title = "",
ncol = 1,
override.aes = list(#shape = v_shape,
linetype = v_linetype,
color = v_colors_colour,
fill = v_colors_fill))) +
geom_vline(xintercept = c(as.numeric(as.Date("2020-03-23")), as.numeric(n_date_last)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed", "twodash"),
color = "#393F3E") +
geom_hline(yintercept = 1, linetype = 3) +
annotation_custom(npi_date) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks
) +
scale_y_continuous(outcome_eng,
breaks = number_ticks(8)#,
#labels = scales::comma
) + # trans = "log"
# labs(title = "COVID-19 effective reproduction number in MCMA, Mexico" #, caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
# plot.caption = element_text(hjust = 0,
# size = caption_size),
axis.text.x = element_text(angle = angle_x_axis,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
legend.position = "bottom",
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
## Interventions ----------------------------------------------------------
# Abbreviation for outcome to save the plot
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Colors
v_names_colours <- c(as.character(levels(df_RTt_INT$Intervention)[-1]))
# Colours
v_colors_colour <- c(#"#659583",
"#008450", "#FF9900", "#B81D13")
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- v_colors_colour
names(v_colors_fill) <- v_names_colours
# # Vector of colors to shape
# v_colors_shape <- c(#8,
# 15, 16, 17, 18)
# names(v_colors_shape) <- v_names_colours
# End
df_RTt_INT_temp <- subset(df_RTt_INT, dates >= n_date_last & proj_type == n_proj_type)
y_end <- max(df_RTt_INT_temp$ub)
y_start <- min(df_RTt_INT_temp$lb)
# For annotation_custom
scenarios_start <- grobTree(textGrob("Policy start date",
x = 0.145,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
# calib_day <- grobTree(textGrob("Last day used for calibration",
# x = 0.675,
# y = 0.97,
# hjust = 0,
# gp = gpar(col = "#393F3E",
# fontsize = annotation_size)))
### Plot: Interventions ---------------------------------------------------
l_plots_S4[[i+1]] <- ggplot(data = df_RTt_INT_temp) +
geom_line(aes(x = dates,
y = value,
colour = Intervention),
size = 0.9) +
geom_ribbon(aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = Intervention),
alpha = 0.4,
colour = NA) +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour, 2)) +
scale_fill_manual(values = rep(v_colors_fill, 2)) +
guides(fill = guide_legend(title = ""),
color = guide_legend(title = "",
ncol = 1,
override.aes = list(color = v_colors_colour,
fill = v_colors_fill #, shape = v_colors_shape
))) +
geom_vline(xintercept = as.numeric(date_NPI_update),
show.legend = TRUE,
size = 0.5,
linetype = "dashed",
color = "#393F3E") +
geom_hline(yintercept = 1, linetype = 3) +
annotation_custom(scenarios_start) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks[20:26]
) +
scale_y_continuous(outcome_eng,
breaks = number_ticks(8)#,
#labels = scales::comma
) + # trans = "log"
# labs(title = "COVID-19 effective reproduction number in MCMA, Mexico" #, caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
# plot.caption = element_text(hjust = 0,
# size = caption_size),
axis.text.x = element_text(angle = 0,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "bottom",
legend.text = element_text(size = 16),
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent")) +
coord_cartesian(ylim = c(y_start*0.90, y_end*1.1))
## Combine two graphs -----------------------------------------------------
plot_figS4 <- ggarrange(l_plots_S4[[i]],
NULL,
l_plots_S4[[i+1]],
ncol = 1,
heights = c(1, 0.1, 1),
labels = c("A","","B"), hjust = 0,vjust = 1.5,label.x = c(0,0,0), label.y = c(1,1,1),
font.label = list(size = 26, face = "bold"))
ggsave(paste0("figs/figs_paper/figS4_Rt_BaseCase_INT_",n_proj_type,".jpg"),
width = 16, height = 18, dpi = 300)
i <- i + 2
}
# FIGURE S6: Cumulative proportion of infections being detected -----------
# Compute proportions
df_outcome_all_detec_inf <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
Outcome == "Cumulative cases" & BaseCase_type == "Holidays" &
intervention_type == "BaseCase" &
dates <= max_date)
df_outcome_all_total_inf <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
Outcome == "Cumulative infections" & BaseCase_type == "Holidays" &
intervention_type == "BaseCase" &
dates <= max_date)
df_outcome_all_detec_inf$Intervention <- "Model-projected"
df_outcome_all_total_inf$Intervention <- "Model-projected"
# Compute proportion
df_prop_DetecInf <- data.frame(df_outcome_all_total_inf[,c("type","BaseCase_type", "Intervention",
"intervention_type", "Outcome", "dates",
"proj_type", "proj_type_label")],
value = df_outcome_all_detec_inf$value/df_outcome_all_total_inf$value
)
df_figS6_DetecInf_prop <- df_prop_DetecInf %>%
group_by(type, BaseCase_type, Intervention, intervention_type, Outcome, dates,
proj_type, proj_type_label) %>%
summarise(mean = mean(value),
median = quantile(value, probs = 0.5, names = FALSE),
sd = sd(value),
lb = quantile(value, probs = 0.025, names = FALSE),
ub = quantile(value, probs = 0.975, names = FALSE)) %>%
rename(value = mean)
# Intervention as factor
df_figS6_DetecInf_prop$Intervention <- as.factor(df_figS6_DetecInf_prop$Intervention)
# Save
save(df_figS6_DetecInf_prop,
file = paste0("figs/figs_paper/data_frames/df_figS6_DetecInf_prop.RData"))
# Abbreviation for outcome to save the plot
abbrev_outcome <- "DetecInf_prop"
# Vector of colors
v_names_colours <- as.character(levels(df_figS6_DetecInf_prop$Intervention))
v_colors_colour <- c("#659583")
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- c("#659583")
names(v_colors_fill) <- v_names_colours
# Line
v_linetype <- c("solid")
names(v_linetype) <- v_names_colours
# Bracket position
y_end <- max(df_figS6_DetecInf_prop$ub)
# For annotation_custom
npi_date <- grobTree(textGrob("Date of NPI implementation",
x = 0.119,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
calib_day <- grobTree(textGrob("Last day used for calibration",
x = 0.675,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
for(n_proj_type in c("SQ")){
ggplot(data = subset(df_figS6_DetecInf_prop, proj_type == n_proj_type)) +
geom_line(aes(x = dates,
y = value,
colour = Intervention),
size = 0.9) +
geom_ribbon(aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = Intervention),
alpha = 0.4,
colour = NA) +
scale_color_manual("", values = v_colors_colour) +
scale_fill_manual("", values = v_colors_fill) +
geom_vline(xintercept = c(as.numeric(as.Date("2020-03-23")), as.numeric(n_date_last)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(npi_date) +
# annotation_custom(calib_day) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks
) +
scale_y_continuous("",
breaks = number_ticks(8),
labels = scales::percent) + # trans = "log"
# labs(title = "COVID-19 total infections in MCMA, Mexico" #, caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
axis.text.x = element_text(angle = angle_x_axis,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "none",
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
# Save plot
ggsave(paste0("figs/figs_paper/figS6_DetecInf_prop_",n_proj_type,".jpg"),
width = 12, height = 9, dpi = 300)
}
# FIGURE S5: Cumulative proportion of population ever been infected -------
# Outcome
outcome <- outcome_eng <- "Recovered prevalence proportion"
# Data.frame
df_figS5_Rec_prop <- df_out_total_prob %>%
ungroup() %>%
filter(state == state_i &
Outcome == outcome_eng & BaseCase_type == "Holidays" &
intervention_type == "BaseCase" &
dates <= max_date)
# Set type as factor
df_figS5_Rec_prop$type <- as.factor(df_figS5_Rec_prop$type)
# Save
save(df_figS5_Rec_prop,
file = paste0("figs/figs_paper/data_frames/df_figS5_Rec_prop.RData"))
# Plot specifications -----------------------------------------------------
# Abbreviation for outcome to save the plot
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Name of colors
v_names_colours <- as.character(levels(df_figS5_Rec_prop$type))
# Colours
v_colors_colour <- c("#a13023")
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- c("#a13023")
names(v_colors_fill) <- v_names_colours
# Line
v_linetype <- c("solid")
names(v_linetype) <- v_names_colours
# Bracket position
y_end <- max(df_figS5_Rec_prop$ub)
# For annotation_custom
npi_date <- grobTree(textGrob("Date of NPI implementation",
x = 0.117,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
calib_day <- grobTree(textGrob("Last day used for calibration",
x = 0.744,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot --------------------------------------------------------------
for(n_proj_type in c("SQ")){
ggplot(data = subset(df_figS5_Rec_prop, proj_type == n_proj_type)) +
geom_line(aes(x = dates,
y = value,
colour = type),
size = 0.9) +
geom_ribbon(aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = type),
alpha = 0.4,
colour = NA) +
scale_color_manual("", values = v_colors_colour) +
scale_fill_manual("", values = v_colors_fill) +
geom_vline(xintercept = c(as.numeric(as.Date("2020-03-23")), as.numeric(n_date_last)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(npi_date) +
# annotation_custom(calib_day) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks
) +
scale_y_continuous("",#outcome_eng,
breaks = number_ticks(8),
labels = scales::percent) + # trans = "log"
# labs(title = "COVID-19 total recovered in MCMA, Mexico" #, caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
# plot.caption = element_text(hjust = 0,
# size = caption_size),
axis.text.x = element_text(angle = angle_x_axis,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "none",
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
# Save plot
ggsave(paste0("figs/figs_paper/figS5_Rec_prop_",n_proj_type,".jpg"),
width = 12, height = 9, dpi = 300)
}
# FIGURE S7: Percentage difference STATUS QUO VS HOLIDAYS -----------------
# Outcomes to be plotted
v_outcome <- c("Incident cases",
"Incident deaths",
"Cumulative cases",
"Cumulative deaths")
# Filter data and compute weekly data
df_CasesDeaths_daily <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
Outcome %in% v_outcome &
intervention_type == "BaseCase" &
dates > n_date_last)
# Replicate status quo value
df_CasesDeaths_daily$BaseCase_value <- 0
df_CasesDeaths_daily$BaseCase_value[df_CasesDeaths_daily$BaseCase_type == "Holidays"] <- df_CasesDeaths_daily$value[df_CasesDeaths_daily$BaseCase_type == "StatusQuo"]
df_CasesDeaths_daily$BaseCase_value[df_CasesDeaths_daily$BaseCase_type == "StatusQuo"] <- df_CasesDeaths_daily$value[df_CasesDeaths_daily$BaseCase_type == "StatusQuo"]
# Compute weekly percentage increases
df_figS7_PercDiff <- df_CasesDeaths_daily %>%
mutate(diffPerc = (value - BaseCase_value)/BaseCase_value) %>%
group_by(type, BaseCase_type, Intervention, intervention_type, Outcome,
dates, proj_type, proj_type_label) %>%
summarise(value = mean(diffPerc),
median = quantile(diffPerc, probs = 0.5, names = FALSE),
sd = sd(diffPerc),
lb = quantile(diffPerc, probs = 0.025, names = FALSE),
ub = quantile(diffPerc, probs = 0.975, names = FALSE)) %>%
filter(BaseCase_type == "Holidays")
# Save
save(df_figS7_PercDiff,
file = paste0("figs/figs_paper/data_frames/df_figS7_PercDiff.RData"))
# Order outcomes
df_figS7_PercDiff$Outcome <- factor(df_figS7_PercDiff$Outcome,
levels = v_outcome,
ordered = TRUE)
# Colors
v_names_colours <- c(as.character(levels(df_figS7_PercDiff$Outcome)))
# Vector of colors to colour
v_colors_colour <- rep(c("#659583","#a13023"),2)
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- v_colors_colour
names(v_colors_fill) <- v_names_colours
# Vector of colors to shape
v_colors_shape <- c(15, 16, 17,18)
names(v_colors_shape) <- v_names_colours
# For annotation_custom
start_holidays <- grobTree(textGrob("Start holidays",
x = 0.12,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
end_holidays <- grobTree(textGrob("End holidays",
x = 0.756,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
for(n_proj_type in c("SQ")){
ggplot(data = subset(df_figS7_PercDiff, proj_type == n_proj_type),
aes(x = dates,
y = value,
colour = Outcome,
ymin = lb,
ymax = ub,
shape = Outcome)) +
geom_point(#position = position_dodge(width = 5),
size = 4) +
geom_errorbar(#position = position_dodge(width = 5),
size = 1.1) +
facet_wrap(~ Outcome, ncol = 2, scales = "free_y") +
# geom_ribbon(aes(x = dates,
# y = mean_percent,
# ymin = lb_percent,
# ymax = ub_percent,
# fill = Intervention),
# alpha = 0.4,
# colour = NA) +
scale_color_manual(values = v_colors_colour) +
scale_fill_manual(values = v_colors_fill) +
scale_shape_manual(values = v_colors_shape) +
guides(shape = guide_legend(title = ""),
color = guide_legend(title = "",
ncol = 1,
override.aes = list(#linetype = v_linetype,
color = v_colors_colour,
# fill = v_colors_fill,
shape = v_colors_shape))) +
geom_vline(xintercept = c(as.Date("2020-12-24"),
as.Date("2021-01-06")),
show.legend = TRUE,
size = 0.5,
linetype = rep(c("dashed", "twodash"),4),
color = "#393F3E") +
# annotation_custom(start_holidays) +
# annotation_custom(end_holidays) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = c(as.Date("2020-12-07"), as.Date("2020-12-14"),
as.Date("2020-12-21"), as.Date("2020-12-28"),
as.Date("2021-01-04"), as.Date("2021-01-11"),
as.Date("2021-01-18"), as.Date("2021-01-25"),
as.Date("2021-02-01"), as.Date("2021-02-08"),
as.Date("2021-02-15"), as.Date("2021-02-22"),
as.Date("2021-03-01"))
# breaks = "week"
) +
scale_y_continuous("Difference from status quo (%)",
breaks = number_ticks(8),
labels = scales::percent
) + # trans = "log"
# labs(# title = "COVID-19 incident cases in MCMA, Mexico",
# caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = caption_size),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
legend.position = "none", # c(0.44,0.5),
legend.text = element_text(size = 20),
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
# Save plot
ggsave(paste0("figs/figs_paper/figS7_PercDiff_SQ_Holidays_",n_proj_type,".jpg"),
width = 21, height = 15, dpi = 300)
}
SC-COSMO/sccosmomcma documentation built on Dec. 18, 2021, 11:56 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
# 05_figures_paper.R
#-------------------------------------------------------------------------#
# This script creates the figures presented in the MCMA paper. #
# #
# Authors: #
# - Valeria Gracia Olvera, MsC #
#-------------------------------------------------------------------------#
rm(list=ls())
# Load packages and set constants -----------------------------------------
# Libraries
library(tidyverse)
library(ggplot2)
library(dplyr)
library(dampack)
library(ggpubr)
library(grid)
# Dates
n_date_projs <- "2020-12-13"
n_date_obs <- "2020-12-13"
n_time_stamp_hosp <- "2021-03-07"
# Load data ---------------------------------------------------------------
# Projections
load(paste0("output/05_projections_probabilistic_all_SQ_MCMA_",n_date_projs,".RData"))
load(paste0("output/05_projections_probabilistic_SQ_MCMA_",n_date_projs,".RData"))
df_out_total_prob_all <- df_out_mex_total_prob_all
df_out_total_prob <- df_out_mex_total_prob
# Observed data
load("data/l_targets_2020-12-13.RData")
# Hospitalization data
load(paste0("data/df_hosp_MCMA_2020-12-21.RData"))
# Plot specifications ----------------------------------------------------
# Names of outcomes
v_outcomes <- c(Hosp_tot = "Total hospitalizations",
CDR_prop = "CDR proportion",
Dcov_tot = "COVID deaths",
DXCumtot = "Cumulative cases",
InfCumtot = "Cumulative infections",
InfCumprop = "Cumulative infections proportion",
DXtot = "Detected cases",
DXDcov_tot = "Cumulative deaths",
NonICU_tot = "Hospitalizations without ventilator",
ICU_tot = "Hospitalizations with ventilator",
DXDIncCov_tot = "Incident deaths",
DXInctot = "Incident cases",
InfInctot = "Incident infections",
Inftot = "Prevalent infections",
Rt = "R effective",
Rec_tot = "Recovered prevalence",
Rec_prop = "Recovered prevalence proportion")
# State
state_i <- unique(df_out_total_prob$state)
# Abbreviation for state
abbrev_state <- unique(df_out_total_prob$s_code)
# Data time stamp
n_time_stamp <- as.Date(unique(df_out_total_prob$time_stamp))
# Last date
n_date_last <- as.Date("2020-12-07")
# Max date
max_date <- as.Date("2021-03-07")
# NPI date
date_NPI <- as.Date("2020-03-23")
# NPI update
date_NPI_update <- as.Date("2021-01-10")
# Holidays dates
n_holidays_start <- as.Date("2020-12-24")
# Text size
text_size <- 24
caption_size <- 20
annotation_size <- 17
bracket_size <- 1.1
label_bracket_size <- 6.8
arrow_size <- 3
# X axis formating
angle_x_axis <- 90
label_x_axis <- ""
# Date breaks
v_dates_breaks <- c(as.Date("2020-02-15"), as.Date("2020-03-01"),
as.Date("2020-03-15"), as.Date("2020-04-01"),
as.Date("2020-04-15"), as.Date("2020-05-01"),
as.Date("2020-05-15"), as.Date("2020-06-01"),
as.Date("2020-06-15"), as.Date("2020-07-01"),
as.Date("2020-07-15"), as.Date("2020-08-01"),
as.Date("2020-08-15"), as.Date("2020-09-01"),
as.Date("2020-09-15"), as.Date("2020-10-01"),
as.Date("2020-10-15"), as.Date("2020-11-01"),
as.Date("2020-11-15"), as.Date("2020-12-01"),
as.Date("2020-12-15"), as.Date("2021-01-01"),
as.Date("2021-01-15"), as.Date("2021-02-01"),
as.Date("2021-02-15"), as.Date("2021-03-01"))
# Set colors
jet.colors <- colorRampPalette(c("black", "#00007F", "blue", "#007FFF",
"cyan", "#7FFF7F", "yellow", "#FF7F00",
"red", "#7F0000"))
color_map <- jet.colors(100)
# Function
# Number of ticks for plots: dampack package
number_ticks <- function(n) {
function(limits) {
pretty(limits, n + 1)
}
}
# Wrangle data ------------------------------------------------------------
# Set vector of interventions and labels
v_interv_names <- c(BaseCase = "Policy A. Physical distancing: status quo; Schooling: not in-person",
IncreaseSD = "Policy B. Physical distancing: +27% compared to status quo; Schooling: not in-person",
IncreaseSDSchoolSD = "Policy C. Physical distancing: +27% compared to status quo; Schooling: in-person",
SchoolSD = "Policy D. Physical distancing: status quo; Schooling: in-person")
# Rename interventions
df_out_total_prob$Intervention <- as.character(df_out_total_prob$Intervention)
df_out_total_prob_all$Intervention <- as.character(df_out_total_prob_all$Intervention)
for(n_interv in names(v_interv_names)){
interv_type <- v_interv_names[which(names(v_interv_names) == n_interv)]
df_out_total_prob$Intervention[df_out_total_prob$intervention_type == n_interv] <- interv_type
df_out_total_prob_all$Intervention[df_out_total_prob_all$intervention_type == n_interv] <- interv_type
}
df_out_total_prob$Intervention <- factor(df_out_total_prob$Intervention,
levels = v_interv_names,
ordered = TRUE)
df_out_total_prob_all$Intervention <- factor(df_out_total_prob_all$Intervention,
levels = v_interv_names,
ordered = TRUE)
# Projection type
v_proj_labels <- c(SQ = "Base-case",
SA = "Lower Covid-19 risk in children")
# Projection type names
df_out_total_prob$proj_type_label <- ""
df_out_total_prob_all$proj_type_label <- ""
for(n_proj_type in c("SQ", "SA")){
df_out_total_prob$proj_type_label[df_out_total_prob$proj_type == n_proj_type] <- v_proj_labels[n_proj_type]
df_out_total_prob_all$proj_type_label[df_out_total_prob_all$proj_type == n_proj_type] <- v_proj_labels[n_proj_type]
}
df_out_total_prob$proj_type_label <- factor(df_out_total_prob$proj_type_label,
levels = v_proj_labels,
ordered = TRUE)
df_out_total_prob_all$proj_type_label <- factor(df_out_total_prob_all$proj_type_label,
levels = v_proj_labels,
ordered = TRUE)
# Create variable with base case labels
v_basecase_names <- c(StatusQuo = "End-of-year Holidays: status quo",
Holidays = "End-of-year Holidays: higher social contacts")
df_out_total_prob$BaseCase <- ""
df_out_total_prob_all$BaseCase <- ""
for(n_basecase in names(v_basecase_names)){
basecase_type <- v_basecase_names[which(names(v_basecase_names) == n_basecase)]
df_out_total_prob$BaseCase[df_out_total_prob$BaseCase_type == n_basecase] <- basecase_type
df_out_total_prob_all$BaseCase[df_out_total_prob_all$BaseCase_type == n_basecase] <- basecase_type
}
df_out_total_prob$BaseCase <- factor(df_out_total_prob$BaseCase,
levels = v_basecase_names,
ordered = TRUE)
df_out_total_prob_all$BaseCase <- factor(df_out_total_prob_all$BaseCase,
levels = v_basecase_names,
ordered = TRUE)
# Rename observed outcomes
df_out_total_prob$Outcome <- as.character(df_out_total_prob$Outcome)
df_out_total_prob$Outcome[df_out_total_prob$Outcome == "Cumulative detected cases"] <- "Cumulative cases"
df_out_total_prob$Outcome[df_out_total_prob$Outcome == "Incident detected cases"] <- "Incident cases"
df_out_total_prob$Outcome[df_out_total_prob$Outcome == "Detected COVID deaths"] <- "Cumulative deaths"
df_out_total_prob$Outcome[df_out_total_prob$Outcome == "Incident COVID19 deaths infections"] <- "Incident deaths"
df_out_total_prob_all$Outcome <- as.character(df_out_total_prob_all$Outcome)
df_out_total_prob_all$Outcome[df_out_total_prob_all$Outcome == "Cumulative detected cases"] <- "Cumulative cases"
df_out_total_prob_all$Outcome[df_out_total_prob_all$Outcome == "Incident detected cases"] <- "Incident cases"
df_out_total_prob_all$Outcome[df_out_total_prob_all$Outcome == "Detected COVID deaths"] <- "Cumulative deaths"
df_out_total_prob_all$Outcome[df_out_total_prob_all$Outcome == "Incident COVID19 deaths infections"] <- "Incident deaths"
## Offset hospitalizations ------------------------------------------------
offset_flag <- TRUE
# Modify data applying an offset
n_date_offset <- "2020-12-07"
# Outcomes of interest
v_hosp_outcomes <- c("Total hospitalizations",
"Hospitalizations without ventilator",
"Hospitalizations with ventilator")
if(offset_flag){
for(n_proj_type in c("SQ")){
for(n_hosp_outcome in c(v_hosp_outcomes)){
denominator <- df_hosp_MCMA$hosp_tot[df_hosp_MCMA$dates == n_date_offset]
offset_hosp_ratio <- mean(df_out_total_prob_all$value[df_out_total_prob_all$dates == n_date_offset &
df_out_total_prob_all$BaseCase_type == "StatusQuo" &
df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$intervention_type == "BaseCase" &
df_out_total_prob_all$Outcome == n_hosp_outcome])/denominator
# Modify data
df_out_total_prob_all$value[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome] <- df_out_total_prob_all$value[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome]/offset_hosp_ratio
df_out_total_prob_all$lb[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome] <- df_out_total_prob_all$lb[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome]/offset_hosp_ratio
df_out_total_prob_all$ub[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome] <- df_out_total_prob_all$ub[df_out_total_prob_all$proj_type == n_proj_type &
df_out_total_prob_all$Outcome == n_hosp_outcome]/offset_hosp_ratio
}
}
}
# Covid-19 cases and deaths observed data
df_covid_obs_data <- rbind(l_targets$cases,
l_targets$cases_inc,
l_targets$deaths,
l_targets$deaths_inc) %>%
ungroup() %>%
filter(state %in% unique(df_out_total_prob$state)) %>%
rename(Outcome = Target,
s_code = abbrev_state,
dates = Date,
sd = se) %>%
select(-country, -county, -var_outcome, -series, -population, -Date0) %>%
mutate(type = "Observed",
BaseCase_type = "Observed",
Intervention = "Observed",
intervention_type = "Observed",
median = NA)
df_covid_obs_data <- df_covid_obs_data[,c("state", "s_code", "type", "BaseCase_type", "Intervention",
"intervention_type", "Outcome", "dates", "value", "median",
"sd", "lb", "ub","time_stamp")]
df_covid_obs_data$Outcome <- as.character(df_covid_obs_data$Outcome)
df_covid_obs_data$Outcome[df_covid_obs_data$Outcome == "Incident confirmed infections"] <- "Incident cases"
df_covid_obs_data$Outcome[df_covid_obs_data$Outcome == "Cumulative confirmed infections"] <- "Cumulative cases"
df_covid_obs_data$Outcome[df_covid_obs_data$Outcome == "Cumulative COVID19 deaths infections"] <-"Cumulative deaths"
df_covid_obs_data$Outcome[df_covid_obs_data$Outcome == "Incident COVID19 deaths infections"] <- "Incident deaths"
# Weekly data
df_covid_obs_data_week <- rbind(l_targets$cases,
l_targets$cases_inc,
l_targets$deaths,
l_targets$deaths_inc) %>%
ungroup() %>%
filter(state %in% unique(df_out_total_prob$state) &
Date <= n_date_last) %>%
rename(Outcome = Target,
s_code = abbrev_state,
dates = Date,
sd = se) %>%
select(-country, -county, -var_outcome, -series, -population, -Date0) %>%
mutate(type = "Observed",
BaseCase_type = "Observed",
Intervention = "Observed",
intervention_type = "Observed",
median = NA) %>%
group_by(state, type, Outcome, DateLast, time_stamp, s_code,
BaseCase_type, Intervention, intervention_type,
week = cut.Date(dates, breaks = "week")) %>%
summarise(value_week = mean(value),
lb_week = mean(lb),
ub_week = mean(ub),
sd_week = mean(sd)) %>%
rename(value = value_week,
lb = lb_week,
ub = ub_week,
sd = sd_week,
dates = week) %>%
mutate(median = NA) %>%
ungroup()
df_covid_obs_data_week <- df_covid_obs_data_week[,c("state", "s_code", "type", "BaseCase_type", "Intervention",
"intervention_type", "Outcome", "dates", "value", "median",
"sd", "lb", "ub","time_stamp")]
df_covid_obs_data_week$Outcome <- as.character(df_covid_obs_data_week$Outcome)
df_covid_obs_data_week$Outcome[df_covid_obs_data_week$Outcome == "Incident confirmed infections"] <- "Incident cases"
df_covid_obs_data_week$Outcome[df_covid_obs_data_week$Outcome == "Cumulative confirmed infections"] <- "Cumulative cases"
df_covid_obs_data_week$Outcome[df_covid_obs_data_week$Outcome == "Cumulative COVID19 deaths infections"] <-"Cumulative deaths"
df_covid_obs_data_week$Outcome[df_covid_obs_data_week$Outcome == "Incident COVID19 deaths infections"] <- "Incident deaths"
# ABSTRACT ----------------------------------------------------------------
# # Interventions
# select_intervention <- c("SchoolSD",
# "IncreaseSD",
# "IncreaseSDSchoolSD",
# "BaseCase")
#
# df_AbstractCases <- df_out_total_prob_all %>%
# filter(dates >= "2020-12-07" &
# # BaseCase_type=="Holidays" &
# intervention_type %in% select_intervention &
# Outcome == "Incident cases") %>%
# group_by(county, type, BaseCase_type, Intervention, intervention_type, Outcome,
# simulation, proj_type, proj_type_label) %>%
# summarise(cum_value = sum(value)) %>%
# group_by(county, type, BaseCase_type, Intervention, intervention_type, Outcome,
# proj_type, proj_type_label) %>%
# summarise(value = round(mean(cum_value/1e6),2),
# sd = round(sd(cum_value/1e6),2),
# lb = round(quantile(cum_value/1e6, probs = 0.025, names = FALSE), 2),
# ub = round(quantile(cum_value/1e6, probs = 0.975, names = FALSE), 2)) %>%
# mutate(time_stamp = n_time_stamp) %>%
# rename(state = county)
#
# df_AbstractHosp <- df_out_total_prob_all %>%
# filter(dates >= "2020-12-07" &
# intervention_type %in% select_intervention &
# Outcome == "Total hospitalizations") %>%
# group_by(county, type, BaseCase_type, Intervention, intervention_type, Outcome,
# proj_type, proj_type_label, dates) %>%
# summarise(mean = round(mean(value),2),
# sd = round(sd(value),2),
# lb = round(quantile(value, probs = 0.025, names = FALSE), 2),
# ub = round(quantile(value, probs = 0.975, names = FALSE), 2)) %>%
# mutate(time_stamp = n_time_stamp) %>%
# rename(state = county)
#
# # First value:
# df_AbstractHosp[df_AbstractHosp$BaseCase_type == "Holidays" &
# df_AbstractHosp$intervention_type == "BaseCase" &
# df_AbstractHosp$mean == max(df_AbstractHosp$mean[df_AbstractHosp$BaseCase_type == "Holidays" &
# df_AbstractHosp$intervention_type == "BaseCase"]),]
# # Second value
# df_AbstractHosp[df_AbstractHosp$BaseCase_type == "StatusQuo" &
# df_AbstractHosp$intervention_type == "IncreaseSDSchoolSD" &
# df_AbstractHosp$mean == max(df_AbstractHosp$mean[df_AbstractHosp$BaseCase_type == "StatusQuo" &
# df_AbstractHosp$intervention_type == "IncreaseSDSchoolSD"]),]
# FIGURE 1: cases and deaths ----------------------------------------------
# Vector of outcomes of interest
v_outcome <- c("Incident cases",
"Incident deaths",
"Cumulative cases",
"Cumulative deaths")
# Wrangle projections
df_outcome_int <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
Outcome %in% v_outcome &
intervention_type == "BaseCase" &
dates <= max_date) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, proj_type, proj_type_label) %>%
summarise(mean = mean(value),
median = quantile(value, probs = 0.5, names = FALSE),
sd = sd(value),
lb = quantile(value, probs = 0.025, names = FALSE),
ub = quantile(value, probs = 0.975, names = FALSE)) %>%
rename(value = mean,
state = county) %>%
mutate(time_stamp = n_time_stamp) %>%
select(state, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, value, lb, ub, time_stamp,
proj_type, proj_type_label) %>%
ungroup()
df_outcome_int$Intervention <- "Model-predicted"
# Observed data
df_outcome_obs <- rbind.data.frame(
subset(df_covid_obs_data_week, Intervention == "Observed") %>%
mutate(BaseCase_type = "Holidays",
BaseCase = v_basecase_names["Holidays"],
proj_type = "SQ",
proj_type_label = "Base-case"),
subset(df_covid_obs_data_week, Intervention == "Observed") %>%
mutate(BaseCase_type = "StatusQuo",
BaseCase = v_basecase_names["StatusQuo"],
proj_type = "SQ",
proj_type_label = "Base-case")) %>%
select(state, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, value, lb, ub, time_stamp,
proj_type, proj_type_label)
df_outcome_obs$Intervention <- as.factor(df_outcome_obs$Intervention)
# Create and save data.frame
df_fig1_CasesDeaths <- rbind.data.frame(df_outcome_int,
df_outcome_obs)
# Save data
save(df_fig1_CasesDeaths,
file = paste0("figs/figs_paper/data_frames/df_fig1_CasesDeaths.RData"))
## Plot specifications ----------------------------------------------------
# Colors
v_names_colours_obs <- as.character(levels(df_fig1_CasesDeaths$Intervention))
v_colors_colour <- rev(c("#a13023","#659583"))
names(v_colors_colour) <- v_names_colours_obs
# Vector of colors to fill
v_colors_fill <- rev(c(NA,"#659583"))
names(v_colors_fill) <- v_names_colours_obs
# Shape
v_shape <- rev(c(16,NA))
names(v_shape) <- v_names_colours_obs
# Line
v_linetype <- rev(c("blank","solid"))
names(v_linetype) <- v_names_colours_obs
# For annotation_custom
npi_date <- grobTree(textGrob("Date of NPI implementation",
x = 0.12,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
calib_day <- grobTree(textGrob("Last day used for calibration",
x = 0.756,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
# Save plots in list
l_plots_fig1 <- vector(mode = "list", length = 4)
i <- 1
for(n_proj_type in c("SQ")){ # n_proj_type = "SA"
for(outcome in v_outcome){
# Abbreviation for outcome to save the plot
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Filter data to outcome and type of data
df_obs_temp <- subset(df_fig1_CasesDeaths,
Outcome == outcome &
Intervention == "Observed" &
proj_type == n_proj_type)
df_int_temp <- subset(df_fig1_CasesDeaths,
Outcome == outcome &
Intervention == "Model-predicted" &
proj_type == n_proj_type)
# Bracket position
y_end <- max(df_obs_temp$ub,df_int_temp$ub)
# Plot
l_plots_fig1[[i]] <- ggplot(data = df_obs_temp) +
geom_point(mapping = aes(x = dates,
y = value,
colour = Intervention,
fill = Intervention),
shape = 16,
size = 3,
alpha = 0.6) +
geom_line(data = df_int_temp,
aes(x = dates,
y = value,
colour = Intervention),
size = 1.1) + #0.9) +
geom_ribbon(data = df_int_temp,
aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = Intervention),
alpha = 0.4,
colour = NA) +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour,2)) +
scale_fill_manual(values = rep(v_colors_fill, 2)) +
guides(fill = guide_legend(title = ""),
color = guide_legend(title = "",
ncol = 2,
override.aes = list(shape = v_shape,
linetype = v_linetype,
color = v_colors_colour,
fill = v_colors_fill))) +
geom_vline(xintercept = c(as.numeric(as.Date("2020-03-23")),
as.numeric(n_date_last)),
show.legend = TRUE,
size = 0.5,
linetype = rep(c("dashed", "twodash"), 2),
color = "#393F3E") +
annotation_custom(npi_date) +
geom_bracket(data = NULL, xmin = n_date_last,
xmax = max_date,
y.position = y_end*1.04,
label = "Projections",
label.size = label_bracket_size,
size = bracket_size) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks
) +
scale_y_continuous(outcome,
breaks = number_ticks(12),
labels = scales::comma) + # trans = "log"
theme(text = element_text(size = text_size),
axis.text.x = element_text(angle = angle_x_axis,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "bottom",
legend.justification = "center",
legend.text = element_text(size = 22),
legend.direction = "horizontal",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent",
size = unit(3, "cm"))) +
coord_cartesian(ylim = c(0, y_end*1.1))
i <- i + 1
}
i <- 1
# Combine graphs
plot_inc <- ggarrange(l_plots_fig1[[i]],
NULL,
l_plots_fig1[[i+1]],
ncol = 1,
common.legend = T,
legend = "none",
heights = c(1,0.05, 1),
labels = c("A","","B"),
hjust = 0,vjust = 1.5,
font.label = list(size = 26, face = "bold"))
# Cumulative
plot_cum <- ggarrange(l_plots_fig1[[i+2]],
NULL,
l_plots_fig1[[i+3]],
ncol = 1,
common.legend = T,
legend = "bottom",
heights = c(1,0.05, 1),
labels = c("C","","D"),
hjust = 0,vjust = 1.5,
font.label = list(size = 26, face = "bold"))
plot_fig1 <- ggarrange(plot_inc,
plot_cum,
ncol = 1)
# Save plot
ggsave(paste0("figs/figs_paper/fig1_CasesDeaths_",n_proj_type,".jpg"),
width = 18, height = 32, dpi = 300)
}
# FIGURE 2: Hospitalizations ----------------------------------------------
# Vector of outcomes of interest
outcome <- "Total hospitalizations"
# Offset flag
offset_flag <- TRUE
# Hosp projected data
df_Hosp_proj <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
intervention_type == "BaseCase" &
Outcome == outcome &
dates <= max_date) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, proj_type, proj_type_label) %>%
summarise(mean = round(mean(value),0),
median = round(quantile(value, probs = 0.5, names = FALSE),0),
sd = round(sd(value),0),
lb = round(quantile(value, probs = 0.025, names = FALSE),0),
ub = round(quantile(value, probs = 0.975, names = FALSE),0)) %>%
rename(value = mean,
state = county) %>%
mutate(time_stamp = n_time_stamp,
cap_tot = NA) %>%
select(state, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, value, lb, ub, cap_tot, time_stamp,
proj_type, proj_type_label)
df_Hosp_proj$Intervention <- "Model-predicted"
# Observed data
df_hosp_tot <- df_hosp_MCMA %>%
filter(dates <= n_date_last)
df_hosp_tot$dates <- as.Date(df_hosp_tot$dates)
df_hosp_tot <- df_hosp_tot[order(df_hosp_tot$dates,decreasing = F),]
df_hosp_tot <- df_hosp_tot %>%
mutate(state = state_i,
type = "Observed",
Outcome = outcome,
Intervention = "Observed",
intervention_type = "Observed",
lb = NA,
ub = NA, time_stamp = as.Date(n_date_obs)
) %>%
filter(dates <= n_time_stamp_hosp) %>%
rename(value = hosp_tot)
df_hosp_tot$Intervention <- as.factor(df_hosp_tot$Intervention)
# Create data.frame with two set of base-case
df_hosp_observed <- rbind.data.frame(df_hosp_tot %>% mutate(BaseCase_type = "Holidays",
BaseCase = v_basecase_names["Holidays"],
proj_type = "SQ",
proj_type_label = "Base-case"),
df_hosp_tot %>% mutate(BaseCase_type = "StatusQuo",
BaseCase = v_basecase_names["StatusQuo"],
proj_type = "SQ",
proj_type_label = "Base-case")) %>%
select(state, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, value, lb, ub, cap_tot, time_stamp,
proj_type, proj_type_label
)
# Bind data.frames
df_fig2_HospTot <- rbind.data.frame(df_Hosp_proj,
df_hosp_observed)
# Save
if(offset_flag){
save(df_fig2_HospTot,
file = paste0("figs/figs_paper/data_frames/df_fig2_HospTot_offset.RData"))
}else{
save(df_fig2_HospTot,
file = paste0("figs/figs_paper/data_frames/df_fig2_HospTot.RData"))
}
## Plot specifications ----------------------------------------------------
# Order interventions
v_interv <- as.character(unique(df_fig2_HospTot$Intervention))
df_fig2_HospTot$Intervention <- ordered(df_fig2_HospTot$Intervention,
rev(v_interv))
# Colors
v_names_colours_obs <- as.character(levels(df_fig2_HospTot$Intervention))
v_colors_colour <- c("#a13023", "#659583")
names(v_colors_colour) <- v_names_colours_obs
# Vector of colors to fill
v_colors_fill <- c("#a13023","#659583")
names(v_colors_fill) <- v_names_colours_obs
# Line
v_linetype <- c("solid","solid")
names(v_linetype) <- v_names_colours_obs
# For annotation_custom
npi_date <- grobTree(textGrob("Date of NPI implementation",
x = 0.12,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
calib_day <- grobTree(textGrob("Last day used for calibration",
x = 0.675,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
# List of plots
l_plot_fig2 <- vector(mode = "list", length = 2)
i <- 1
for(n_proj_type in c("SQ")){
# Abbreviation for outcome to save the plot
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Filter outcome of interest
df_obs_temp <- subset(df_fig2_HospTot,
Intervention == "Observed" &
proj_type == n_proj_type)
df_int_temp <- subset(df_fig2_HospTot,
Intervention == "Model-predicted" &
proj_type == n_proj_type)
# Bracket position
y_end <- max(df_obs_temp$value,df_int_temp$ub)
# Plot
l_plot_fig2[[i]] <- ggplot(data = df_obs_temp) +
geom_area(position = "stack",
aes(x = dates,
y = value,
colour = Intervention,
fill = Intervention),
alpha = 0.4) +
geom_line(data = df_int_temp,#subset(df_outcome_int, dates >= max(df_hosp_tot$dates)),
aes(x = dates,
y = value,
colour = Intervention),
linetype = "solid") +
geom_ribbon(data = df_int_temp, #subset(df_outcome_int, dates >= max(df_hosp_tot$dates)),
aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = Intervention),
alpha = 0.4,
colour = NA) +
geom_line(data = df_obs_temp,
aes(x = dates,
y = cap_tot),
size = 1.2,
color = "#393F3E") +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour, 2)) +
scale_fill_manual(values = rep(v_colors_fill, 2)) +
# scale_color_manual(values = v_colors_colour) +
# scale_fill_manual(values = v_colors_fill) +
guides(fill = guide_legend(title = "",
reverse = T),
color = guide_legend(title = "",
ncol = 2,
override.aes = list(color = v_colors_colour,
fill = v_colors_fill),
reverse = T)) +
geom_vline(xintercept = c(as.numeric(as.Date("2020-03-23")), as.numeric(n_date_last)),
show.legend = TRUE,
size = 0.5,
linetype = rep(c("dashed", "twodash"), 2),
# linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(npi_date) +
# annotation_custom(calib_day) +
annotate("text",
label = paste0("Current capacity: ",
last(df_obs_temp$cap_tot)," beds"),
x = as.Date("2020-09-13"),
y = last(df_obs_temp$cap_tot) + 2000,
size = 6,
colour = "#393F3E") +
geom_bracket(data = NULL, xmin = n_date_last,
xmax = max_date,
y.position = y_end*1.04,
label = "Projections",
label.size = label_bracket_size,
size = bracket_size) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks,
) +
scale_y_continuous(outcome,
breaks = number_ticks(8),
labels = scales::comma) + # trans = "log"
# labs(title = "COVID-19 hospitalizations with ventilator in MCMA, Mexico" # , caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = caption_size),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "bottom",
legend.justification = "center",
legend.direction = "vertical",
legend.text = element_text(size = 22),
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent",
size = unit(3, "cm"))) +
coord_cartesian(ylim = c(0, y_end*1.1))
# Save plot
if(offset_flag){
ggsave(paste0("figs/figs_paper/fig2_HospTot_offset_",n_proj_type,".jpg"),
width = 16, height = 9, dpi = 300)
}else{
ggsave(paste0("figs/figs_paper/fig2_HospTot_",n_proj_type,".jpg"),
width = 16, height = 9, dpi = 300)
}
i <- i + 1
}
# FIGURE 3: interventions cases and deaths --------------------------------
# Outcomes
v_outcome <- c("Incident deaths",
"Incident cases")
# Interventions
select_intervention <- c("SchoolSD",
"IncreaseSD",
"IncreaseSDSchoolSD",
"BaseCase")
# Data.frame
df_fig3_CasesDeaths_interv <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
Outcome %in% v_outcome &
intervention_type %in% select_intervention &
dates > n_date_last &
dates <= max_date) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, proj_type, proj_type_label) %>%
summarise(mean = mean(value),
median = quantile(value, probs = 0.5, names = FALSE),
sd = sd(value),
lb = quantile(value, probs = 0.025, names = FALSE),
ub = quantile(value, probs = 0.975, names = FALSE)) %>%
rename(value = mean)
# Order outcomes
v_outcom <- as.character(unique(df_fig3_CasesDeaths_interv$Outcome))
df_fig3_CasesDeaths_interv$Outcome <- ordered(df_fig3_CasesDeaths_interv$Outcome,
v_outcom[c(2,1)])
# Save data.frame
save(df_fig3_CasesDeaths_interv,
file = paste0("figs/figs_paper/data_frames/df_fig3_CasesDeaths_interv.RData"))
## Plot specifications ----------------------------------------------------
# Colors
v_names_colours <- c(as.character(levels(df_fig3_CasesDeaths_interv$Intervention)))
v_colors_colour <- c("#0F57CA", "#008450", "#FF9900", "#B81D13")
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- v_colors_colour
names(v_colors_fill) <- v_names_colours
# Vector of colors to shape
v_colors_shape <- c(15, 16, 17, 18)
names(v_colors_shape) <- v_names_colours
# Bracket position
y_end <- max(df_fig3_CasesDeaths_interv$value)
NPI_update <- grobTree(textGrob("Policy start date",
x = 0.14,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
# List of plots
i <- 1
l_plots_fig3 <- vector(mode = "list", length = 4)
for(n_proj_type in c("SQ")){
# Plot incident cases
l_plots_fig3[[i]] <- ggplot(
data = subset(df_fig3_CasesDeaths_interv, Outcome == "Incident cases" &
proj_type == n_proj_type),
aes(x = dates,
y = value,
colour = Intervention,
shape = Intervention)) +
# geom_bar(stat = "identity", position = position_dodge(width = 5),
# width = 3) +
geom_line(size = 0.9) +
# geom_point(size = 3) +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour, 2)) +
scale_shape_manual(values = rep(v_colors_shape, 2)) +
geom_vline(xintercept = c(as.numeric(date_NPI_update)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed"),
# linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(NPI_update) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = "week") +
scale_y_continuous("Cases",#"Difference to case base scenario",
breaks = number_ticks(8),
# labels = scales::percent
labels = scales::comma) + # trans = "log"
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = caption_size+2),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
legend.position = "bottom", # c(0.44,0.5),
legend.text = element_text(size = 20),
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
# Plot incident deaths
l_plots_fig3[[i+1]] <- ggplot(
data = subset(df_fig3_CasesDeaths_interv, Outcome == "Incident deaths" &
proj_type == n_proj_type),
aes(x = dates,
y = value,
colour = Intervention,
shape = Intervention)) +
geom_line(size = 0.9) +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour, 2)) +
scale_shape_manual(values = rep(v_colors_shape, 2)) +
geom_vline(xintercept = c(as.numeric(date_NPI_update)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed"),
# linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(NPI_update) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = "week") +
scale_y_continuous("Deaths",#"Difference to case base scenario",
breaks = number_ticks(8),
# labels = scales::percent
labels = scales::comma) + # trans = "log"
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = caption_size+2),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
legend.position = "bottom", # c(0.44,0.5),
legend.text = element_text(size = 20),
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
## Combine graphs ---------------------------------------------------------
plot_fig3 <- ggarrange(l_plots_fig3[[i]],
NULL,
l_plots_fig3[[i+1]],
ncol = 1,
common.legend = T,
legend = "bottom", heights = c(1, 0.05, 1),
labels = c("A","","B"), hjust = 0,vjust = 1.5,
font.label = list(size = 26, face = "bold"))
# Save plot
ggsave(paste0("figs/figs_paper/fig3_CasesDeaths_interv_",n_proj_type,".jpg"),
width = 16, height = 18, dpi = 300)
i <- i + 2
}
# FIGURE 4: interventions total hospitalizations --------------------------
# Interventions
select_intervention <- c("SchoolSD",
"IncreaseSD",
"IncreaseSDSchoolSD",
"BaseCase")
# Offset flag
offset_flag <- TRUE
# Data.frame
df_hosp_all_INT_fig4 <- df_out_total_prob_all %>%
filter(Outcome == "Total hospitalizations" &
intervention_type %in% select_intervention)
df_hosp_all_INT_fig4 <- df_hosp_all_INT_fig4[order(df_hosp_all_INT_fig4$dates, decreasing = F),]
# Observed data
df_hosp_fig4 <- df_hosp_MCMA %>%
filter(dates <= n_date_last)
df_hosp_fig4$dates <- as.Date(df_hosp_fig4$dates)
df_hosp_fig4 <- df_hosp_fig4[order(df_hosp_fig4$dates,decreasing = F),]
df_hosp_fig4 <- df_hosp_fig4 %>%
mutate(Intervention = "Observed") %>%
filter(dates <= n_date_last)
# Data.frame
df_fig4_HospTot_interv <- df_hosp_all_INT_fig4 %>%
filter(county == state_i &
intervention_type %in% select_intervention) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, proj_type, proj_type_label) %>%
summarise(mean = round(mean(value),0),
median = round(quantile(value, probs = 0.5, names = FALSE),0),
sd = sd(value),
lb = round(quantile(value, probs = 0.025, names = FALSE),0),
ub = round(quantile(value, probs = 0.975, names = FALSE),0)) %>%
rename(value = mean,
state = county
) %>%
filter(dates >= n_date_last) %>%
mutate(cap_tot = unique(df_hosp_observed$cap_tot[df_hosp_observed$dates==n_date_last]),
time_stamp = n_time_stamp)
df_fig4_HospTot_interv$dates <- as.Date(df_fig4_HospTot_interv$dates)
# Save
if(offset_flag){
save(df_fig4_HospTot_interv,
file = paste0("figs/figs_paper/data_frames/df_fig4_HospTot_interv_offset.RData"))
}else{
save(df_fig4_HospTot_interv,
file = paste0("figs/figs_paper/data_frames/df_fig4_HospTot_interv.RData"))
}
## Plot specifications ----------------------------------------------------
# Colors
v_names_colours <- c(as.character(levels(df_fig4_HospTot_interv$Intervention)))
v_colors <- c("#0F57CA", "#008450", "#FF9900", "#B81D13")
# Vector of colors to colour
v_colors_colour <- v_colors
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- v_colors
names(v_colors_fill) <- v_names_colours
# Vector of colors to shape
v_colors_shape <- c(15, 16, 17, 18)
names(v_colors_shape) <- v_names_colours
# Annotations
NPI_update <- grobTree(textGrob("Policy start date",
x = 0.14,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
# List of plots
l_plot_fig4 <- vector(mode = "list", length = 2)
i <- 1
for(n_proj_type in c("SQ")){
# Plot
l_plot_fig4[[i]] <- ggplot(
data = subset(df_fig4_HospTot_interv, proj_type == n_proj_type),
aes(x = dates,
y = value,
colour = Intervention)) +
# geom_bar(stat = "identity", position = position_dodge(width = 5),
# width = 3) +
geom_line(size = 0.9) +
# geom_point(size = 3) +
geom_line(aes(x = dates,
y = cap_tot),
size = 1.2,
color = "#393F3E" #, linetype = "twodash"
) +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour, 2)) +
# scale_shape_manual(values = rep(v_colors_shape, 2)) +
geom_vline(xintercept = c(as.numeric(date_NPI_update)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed"),
# linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(NPI_update) +
annotate("text",
label = paste0("Current capacity: ",
unique(df_hosp_observed$cap_tot[df_hosp_observed$dates==n_date_last])," beds"),
x = as.Date("2020-12-27"),
y = unique(df_hosp_observed$cap_tot[df_hosp_observed$dates==n_date_last]) + 1000,
size = 6,
colour = "#393F3E") +
# scale_fill_manual(values = rep(v_colors_fill, 2)) +
# scale_color_manual(values = v_colors_colour) +
# scale_fill_manual(values = v_colors_fill) +
# guides(fill = guide_legend(title = ""),
# color = guide_legend(title = "",
# ncol = 1,
# override.aes = list(color = v_colors_colour,
# fill = v_colors_fill))) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = "week") +
scale_y_continuous("Total hospitalizations",
breaks = number_ticks(8),
# labels = scales::percent
labels = scales::comma) + # trans = "log"
# labs(# title = "COVID-19 incident cases in MCMA, Mexico",
# caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = caption_size),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
legend.position = "bottom", # c(0.44,0.5),
legend.text = element_text(size = 20),
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
# Save plot
if(offset_flag){
ggsave(paste0("figs/figs_paper/fig4_HospTot_interv_offset_",n_proj_type,".jpg"),
width = 16, height = 9, dpi = 300)
}else{
ggsave(paste0("figs/figs_paper/fig4_HospTot_interv_",n_proj_type,".jpg"),
width = 16, height = 9, dpi = 300)
}
i <- i + 1
}
# FIGURE 5: exceed hosp capacity ------------------------------------------
# Interventions
select_intervention <- c("SchoolSD",
"IncreaseSD",
"IncreaseSDSchoolSD",
"BaseCase")
# Vector of outcomes of interest
outcome <- "Total hospitalizations"
# Hospital capacity
v_hosp_capacity <- c(Hosp_tot = 9667,
NonICU_tot = 7008,
ICU_tot = 2659)
# Interventions data
df_fig5_ExceedHospCap <- df_out_total_prob_all %>%
filter(Outcome == outcome &
intervention_type %in% select_intervention &
dates >= n_date_last) %>%
mutate(cap_tot = df_hosp_MCMA$cap_tot[df_hosp_MCMA$dates == n_date_last]) %>%
mutate(exceed_cap = ifelse(value > cap_tot, 1, 0)) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention, intervention_type,
Outcome, dates, proj_type, proj_type_label) %>%
mutate(exceed_cap_count = sum(exceed_cap),
exceed_cap_prop = exceed_cap_count*100/1000) %>%
summarise(value = mean(exceed_cap_prop),
median = quantile(exceed_cap_prop, probs = 0.5, names = FALSE),
sd = sd(exceed_cap_prop),
lb = quantile(exceed_cap_prop, probs = 0.025, names = FALSE),
ub = quantile(exceed_cap_prop, probs = 0.975, names = FALSE)) %>%
rename(state = county)
# Rename intervention_label
df_fig5_ExceedHospCap$intervention_label <- ""
for(n_interv in names(v_interv_labels)){
df_fig5_ExceedHospCap$intervention_label[df_fig5_ExceedHospCap$intervention_type == n_interv] <- v_interv_labels[n_interv]
}
df_fig5_ExceedHospCap$intervention_label <- factor(df_fig5_ExceedHospCap$intervention_label,
levels = v_interv_labels,
ordered = TRUE)
# Save
if(offset_flag){
save(df_fig5_ExceedHospCap,
file = paste0("figs/figs_paper/data_frames/df_fig5_ExceedHospCap_",abbrev_outcome,"_offset.RData"))
}else{
save(df_fig5_ExceedHospCap,
file = paste0("figs/figs_paper/data_frames/df_fig5_ExceedHospCap_",abbrev_outcome,".RData"))
}
# Plot specifications -----------------------------------------------------
# Caption
# Set vector of interventions and labels
v_interv_labels <- c(BaseCase = "Policy A",
IncreaseSD = "Policy B",
IncreaseSDSchoolSD = "Policy C",
SchoolSD = "Policy D"
)
plot_caption <- ""
for(intv in unique(df_fig5_ExceedHospCap$Intervention)){
plot_caption <- paste0(plot_caption, paste0(intv,"\n"))
}
## Plot -------------------------------------------------------------------
# List of plots
l_plots_fig5 <- vector(mode = "list", length = 2)
i <- 1
for(n_proj_type in c("SQ")){
# Abbrev outcome
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Plot
l_plots_fig5[[i]] <- ggplot(data = subset(df_fig5_ExceedHospCap, proj_type == n_proj_type),
aes(x = dates,
y = intervention_label,
fill = value)) +
geom_raster() +
labs(y = "", fill = "Probability of\nexceeding hospital\ncapacity (%)") +
scale_fill_gradientn(colours = color_map,
breaks = c(0.0, 25.0, 50.0, 75.0, 100.0),
labels = c("0", "25", "50", "75", "100"),
limits = c(0,100)) + # Hawre's jet map
facet_wrap( ~ BaseCase, ncol = 1) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
# breaks = seq.Date(from = first(df_hosp_all_INT_daily$week),
# to = last(df_hosp_all_INT_daily$week),
# by = "week")
breaks = "week") +
labs(caption = plot_caption,
title = outcome) +
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = 19),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "right",
legend.title = element_text(size = 18),
legend.justification = "center",
legend.direction = "vertical",
legend.key.width = unit(0.9, "cm"),
legend.key.height = unit(1.2, "cm"),
legend.title.align = 0,
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
if(offset_flag){
ggsave(paste0("figs/figs_paper/fig5_ExceedHospCap_",abbrev_outcome,"_",n_proj_type,"_offset.jpg"),
width = 16, height = 11, dpi = 300)
}else{
ggsave(paste0("figs/figs_paper/fig5_ExceedHospCap_",abbrev_outcome,"_",n_proj_type,".jpg"),
width = 16, height = 11, dpi = 300)
}
i <- i + 1
}
# First value in abstract:
df_fig5_ExceedHospCap[df_fig5_ExceedHospCap$BaseCase_type == "Holidays" &
df_fig5_ExceedHospCap$intervention_type == "BaseCase" &
df_fig5_ExceedHospCap$dates == "2021-01-25",]
# Second value in abstract:
df_fig5_ExceedHospCap[df_fig5_ExceedHospCap$BaseCase_type == "StatusQuo" &
df_fig5_ExceedHospCap$intervention_type == "IncreaseSDSchoolSD" &
df_fig5_ExceedHospCap$dates == "2021-01-19",]
# Table S2: cases and deaths ----------------------------------------------
v_outcome <- c("Incident cases",
"Cumulative cases",
"Incident deaths",
"Cumulative deaths")
# Interventions
select_intervention <- c("SchoolSD",
"IncreaseSD",
"IncreaseSDSchoolSD",
"BaseCase")
# Intervention data
df_tableS2 <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
intervention_type %in% select_intervention &
Outcome %in% v_outcome &
dates == max_date) %>%
group_by(county, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, proj_type, proj_type_label) %>%
summarise(mean = round(mean(value),0),
median = quantile(value, probs = 0.5, names = FALSE),
sd = round(sd(value),0),
lb = round(quantile(value, probs = 0.025, names = FALSE),0),
ub = round(quantile(value, probs = 0.975, names = FALSE),0)) %>%
rename(value = mean,
state = county) %>%
mutate(time_stamp = n_time_stamp) %>%
select(state, type, BaseCase, BaseCase_type, Intervention,
intervention_type, Outcome, dates, value, lb, ub, time_stamp,
proj_type, proj_type_label)
# Save
save(df_tableS2, file = "figs/figs_paper/data_frames/df_tableS2.RData")
# Figure S4: Effective Reproduction Number --------------------------------
## Status quo -------------------------------------------------------------
# Outcome
outcome <- "R effective"
outcome_eng <- "R effective"
# Data.frame
df_outcome_Rt <- df_out_total_prob %>%
ungroup() %>%
filter(state == state_i &
Outcome == outcome_eng &
dates <= max_date)
# Create data.frame for BaseCase
df_Rt_BaseCase <- subset(df_outcome_Rt, intervention_type == "BaseCase")
df_Rt_BaseCase$Intervention <- "Model-projected"
# Create data.frame for Interventions
df_RTt_INT <- subset(df_outcome_Rt, intervention_type %in% c("SchoolSD", "IncreaseSD","IncreaseSDSchoolSD"))
# Save data.frame
df_figS4_Rt <- rbind.data.frame(df_Rt_BaseCase,
df_RTt_INT)
save(df_figS4_Rt,
file = paste0("figs/figs_paper/data_frames/df_figS4_Rt.RData"))
# List of plots
l_plots_S4 <- vector(mode = "list", length = 4)
i <- 1
for(n_proj_type in c("SQ")){
### Plot: SQ --------------------------------------------------------------
# Abbreviation for outcome to save the plot
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Name of colors
v_names_colours <- as.character(levels(df_Rt_BaseCase$BaseCase))
# Colors
v_colors_colour <- c("#659583", "#a13023" )
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- v_colors_colour
names(v_colors_fill) <- v_names_colours
# Line
v_linetype <- c("solid", "solid")
names(v_linetype) <- v_names_colours
# Bracket position
y_end <- max(df_Rt_BaseCase$ub)
# For annotation_custom
npi_date <- grobTree(textGrob("Date of NPI implementation",
x = 0.115,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
calib_day <- grobTree(textGrob("Last day used for calibration",
x = 0.675,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
# Plot
l_plots_S4[[i]] <- ggplot(data = subset(df_Rt_BaseCase, proj_type == n_proj_type)) +
geom_line(aes(x = dates,
y = value,
colour = BaseCase),
size = 0.9) +
geom_ribbon(aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = BaseCase),
alpha = 0.4,
colour = NA) +
scale_color_manual(values = v_colors_colour) +
scale_fill_manual(values = v_colors_fill) +
guides(fill = guide_legend(title = ""),
color = guide_legend(title = "",
ncol = 1,
override.aes = list(#shape = v_shape,
linetype = v_linetype,
color = v_colors_colour,
fill = v_colors_fill))) +
geom_vline(xintercept = c(as.numeric(as.Date("2020-03-23")), as.numeric(n_date_last)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed", "twodash"),
color = "#393F3E") +
geom_hline(yintercept = 1, linetype = 3) +
annotation_custom(npi_date) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks
) +
scale_y_continuous(outcome_eng,
breaks = number_ticks(8)#,
#labels = scales::comma
) + # trans = "log"
# labs(title = "COVID-19 effective reproduction number in MCMA, Mexico" #, caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
# plot.caption = element_text(hjust = 0,
# size = caption_size),
axis.text.x = element_text(angle = angle_x_axis,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
legend.position = "bottom",
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
## Interventions ----------------------------------------------------------
# Abbreviation for outcome to save the plot
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Colors
v_names_colours <- c(as.character(levels(df_RTt_INT$Intervention)[-1]))
# Colours
v_colors_colour <- c(#"#659583",
"#008450", "#FF9900", "#B81D13")
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- v_colors_colour
names(v_colors_fill) <- v_names_colours
# # Vector of colors to shape
# v_colors_shape <- c(#8,
# 15, 16, 17, 18)
# names(v_colors_shape) <- v_names_colours
# End
df_RTt_INT_temp <- subset(df_RTt_INT, dates >= n_date_last & proj_type == n_proj_type)
y_end <- max(df_RTt_INT_temp$ub)
y_start <- min(df_RTt_INT_temp$lb)
# For annotation_custom
scenarios_start <- grobTree(textGrob("Policy start date",
x = 0.145,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
# calib_day <- grobTree(textGrob("Last day used for calibration",
# x = 0.675,
# y = 0.97,
# hjust = 0,
# gp = gpar(col = "#393F3E",
# fontsize = annotation_size)))
### Plot: Interventions ---------------------------------------------------
l_plots_S4[[i+1]] <- ggplot(data = df_RTt_INT_temp) +
geom_line(aes(x = dates,
y = value,
colour = Intervention),
size = 0.9) +
geom_ribbon(aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = Intervention),
alpha = 0.4,
colour = NA) +
facet_wrap(~BaseCase, ncol = 2) +
scale_color_manual(values = rep(v_colors_colour, 2)) +
scale_fill_manual(values = rep(v_colors_fill, 2)) +
guides(fill = guide_legend(title = ""),
color = guide_legend(title = "",
ncol = 1,
override.aes = list(color = v_colors_colour,
fill = v_colors_fill #, shape = v_colors_shape
))) +
geom_vline(xintercept = as.numeric(date_NPI_update),
show.legend = TRUE,
size = 0.5,
linetype = "dashed",
color = "#393F3E") +
geom_hline(yintercept = 1, linetype = 3) +
annotation_custom(scenarios_start) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks[20:26]
) +
scale_y_continuous(outcome_eng,
breaks = number_ticks(8)#,
#labels = scales::comma
) + # trans = "log"
# labs(title = "COVID-19 effective reproduction number in MCMA, Mexico" #, caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
# plot.caption = element_text(hjust = 0,
# size = caption_size),
axis.text.x = element_text(angle = 0,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "bottom",
legend.text = element_text(size = 16),
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent")) +
coord_cartesian(ylim = c(y_start*0.90, y_end*1.1))
## Combine two graphs -----------------------------------------------------
plot_figS4 <- ggarrange(l_plots_S4[[i]],
NULL,
l_plots_S4[[i+1]],
ncol = 1,
heights = c(1, 0.1, 1),
labels = c("A","","B"), hjust = 0,vjust = 1.5,label.x = c(0,0,0), label.y = c(1,1,1),
font.label = list(size = 26, face = "bold"))
ggsave(paste0("figs/figs_paper/figS4_Rt_BaseCase_INT_",n_proj_type,".jpg"),
width = 16, height = 18, dpi = 300)
i <- i + 2
}
# FIGURE S6: Cumulative proportion of infections being detected -----------
# Compute proportions
df_outcome_all_detec_inf <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
Outcome == "Cumulative cases" & BaseCase_type == "Holidays" &
intervention_type == "BaseCase" &
dates <= max_date)
df_outcome_all_total_inf <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
Outcome == "Cumulative infections" & BaseCase_type == "Holidays" &
intervention_type == "BaseCase" &
dates <= max_date)
df_outcome_all_detec_inf$Intervention <- "Model-projected"
df_outcome_all_total_inf$Intervention <- "Model-projected"
# Compute proportion
df_prop_DetecInf <- data.frame(df_outcome_all_total_inf[,c("type","BaseCase_type", "Intervention",
"intervention_type", "Outcome", "dates",
"proj_type", "proj_type_label")],
value = df_outcome_all_detec_inf$value/df_outcome_all_total_inf$value
)
df_figS6_DetecInf_prop <- df_prop_DetecInf %>%
group_by(type, BaseCase_type, Intervention, intervention_type, Outcome, dates,
proj_type, proj_type_label) %>%
summarise(mean = mean(value),
median = quantile(value, probs = 0.5, names = FALSE),
sd = sd(value),
lb = quantile(value, probs = 0.025, names = FALSE),
ub = quantile(value, probs = 0.975, names = FALSE)) %>%
rename(value = mean)
# Intervention as factor
df_figS6_DetecInf_prop$Intervention <- as.factor(df_figS6_DetecInf_prop$Intervention)
# Save
save(df_figS6_DetecInf_prop,
file = paste0("figs/figs_paper/data_frames/df_figS6_DetecInf_prop.RData"))
# Abbreviation for outcome to save the plot
abbrev_outcome <- "DetecInf_prop"
# Vector of colors
v_names_colours <- as.character(levels(df_figS6_DetecInf_prop$Intervention))
v_colors_colour <- c("#659583")
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- c("#659583")
names(v_colors_fill) <- v_names_colours
# Line
v_linetype <- c("solid")
names(v_linetype) <- v_names_colours
# Bracket position
y_end <- max(df_figS6_DetecInf_prop$ub)
# For annotation_custom
npi_date <- grobTree(textGrob("Date of NPI implementation",
x = 0.119,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
calib_day <- grobTree(textGrob("Last day used for calibration",
x = 0.675,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
for(n_proj_type in c("SQ")){
ggplot(data = subset(df_figS6_DetecInf_prop, proj_type == n_proj_type)) +
geom_line(aes(x = dates,
y = value,
colour = Intervention),
size = 0.9) +
geom_ribbon(aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = Intervention),
alpha = 0.4,
colour = NA) +
scale_color_manual("", values = v_colors_colour) +
scale_fill_manual("", values = v_colors_fill) +
geom_vline(xintercept = c(as.numeric(as.Date("2020-03-23")), as.numeric(n_date_last)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(npi_date) +
# annotation_custom(calib_day) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks
) +
scale_y_continuous("",
breaks = number_ticks(8),
labels = scales::percent) + # trans = "log"
# labs(title = "COVID-19 total infections in MCMA, Mexico" #, caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
axis.text.x = element_text(angle = angle_x_axis,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "none",
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
# Save plot
ggsave(paste0("figs/figs_paper/figS6_DetecInf_prop_",n_proj_type,".jpg"),
width = 12, height = 9, dpi = 300)
}
# FIGURE S5: Cumulative proportion of population ever been infected -------
# Outcome
outcome <- outcome_eng <- "Recovered prevalence proportion"
# Data.frame
df_figS5_Rec_prop <- df_out_total_prob %>%
ungroup() %>%
filter(state == state_i &
Outcome == outcome_eng & BaseCase_type == "Holidays" &
intervention_type == "BaseCase" &
dates <= max_date)
# Set type as factor
df_figS5_Rec_prop$type <- as.factor(df_figS5_Rec_prop$type)
# Save
save(df_figS5_Rec_prop,
file = paste0("figs/figs_paper/data_frames/df_figS5_Rec_prop.RData"))
# Plot specifications -----------------------------------------------------
# Abbreviation for outcome to save the plot
abbrev_outcome <- names(v_outcomes)[which(v_outcomes == outcome)]
# Name of colors
v_names_colours <- as.character(levels(df_figS5_Rec_prop$type))
# Colours
v_colors_colour <- c("#a13023")
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- c("#a13023")
names(v_colors_fill) <- v_names_colours
# Line
v_linetype <- c("solid")
names(v_linetype) <- v_names_colours
# Bracket position
y_end <- max(df_figS5_Rec_prop$ub)
# For annotation_custom
npi_date <- grobTree(textGrob("Date of NPI implementation",
x = 0.117,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
calib_day <- grobTree(textGrob("Last day used for calibration",
x = 0.744,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot --------------------------------------------------------------
for(n_proj_type in c("SQ")){
ggplot(data = subset(df_figS5_Rec_prop, proj_type == n_proj_type)) +
geom_line(aes(x = dates,
y = value,
colour = type),
size = 0.9) +
geom_ribbon(aes(x = dates,
y = value,
ymin = lb,
ymax = ub,
fill = type),
alpha = 0.4,
colour = NA) +
scale_color_manual("", values = v_colors_colour) +
scale_fill_manual("", values = v_colors_fill) +
geom_vline(xintercept = c(as.numeric(as.Date("2020-03-23")), as.numeric(n_date_last)),
show.legend = TRUE,
size = 0.5,
linetype = c("dashed", "twodash"),
color = "#393F3E") +
annotation_custom(npi_date) +
# annotation_custom(calib_day) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = v_dates_breaks
) +
scale_y_continuous("",#outcome_eng,
breaks = number_ticks(8),
labels = scales::percent) + # trans = "log"
# labs(title = "COVID-19 total recovered in MCMA, Mexico" #, caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
# plot.caption = element_text(hjust = 0,
# size = caption_size),
axis.text.x = element_text(angle = angle_x_axis,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
legend.position = "none",
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
# Save plot
ggsave(paste0("figs/figs_paper/figS5_Rec_prop_",n_proj_type,".jpg"),
width = 12, height = 9, dpi = 300)
}
# FIGURE S7: Percentage difference STATUS QUO VS HOLIDAYS -----------------
# Outcomes to be plotted
v_outcome <- c("Incident cases",
"Incident deaths",
"Cumulative cases",
"Cumulative deaths")
# Filter data and compute weekly data
df_CasesDeaths_daily <- df_out_total_prob_all %>%
ungroup() %>%
filter(county == state_i &
Outcome %in% v_outcome &
intervention_type == "BaseCase" &
dates > n_date_last)
# Replicate status quo value
df_CasesDeaths_daily$BaseCase_value <- 0
df_CasesDeaths_daily$BaseCase_value[df_CasesDeaths_daily$BaseCase_type == "Holidays"] <- df_CasesDeaths_daily$value[df_CasesDeaths_daily$BaseCase_type == "StatusQuo"]
df_CasesDeaths_daily$BaseCase_value[df_CasesDeaths_daily$BaseCase_type == "StatusQuo"] <- df_CasesDeaths_daily$value[df_CasesDeaths_daily$BaseCase_type == "StatusQuo"]
# Compute weekly percentage increases
df_figS7_PercDiff <- df_CasesDeaths_daily %>%
mutate(diffPerc = (value - BaseCase_value)/BaseCase_value) %>%
group_by(type, BaseCase_type, Intervention, intervention_type, Outcome,
dates, proj_type, proj_type_label) %>%
summarise(value = mean(diffPerc),
median = quantile(diffPerc, probs = 0.5, names = FALSE),
sd = sd(diffPerc),
lb = quantile(diffPerc, probs = 0.025, names = FALSE),
ub = quantile(diffPerc, probs = 0.975, names = FALSE)) %>%
filter(BaseCase_type == "Holidays")
# Save
save(df_figS7_PercDiff,
file = paste0("figs/figs_paper/data_frames/df_figS7_PercDiff.RData"))
# Order outcomes
df_figS7_PercDiff$Outcome <- factor(df_figS7_PercDiff$Outcome,
levels = v_outcome,
ordered = TRUE)
# Colors
v_names_colours <- c(as.character(levels(df_figS7_PercDiff$Outcome)))
# Vector of colors to colour
v_colors_colour <- rep(c("#659583","#a13023"),2)
names(v_colors_colour) <- v_names_colours
# Vector of colors to fill
v_colors_fill <- v_colors_colour
names(v_colors_fill) <- v_names_colours
# Vector of colors to shape
v_colors_shape <- c(15, 16, 17,18)
names(v_colors_shape) <- v_names_colours
# For annotation_custom
start_holidays <- grobTree(textGrob("Start holidays",
x = 0.12,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
end_holidays <- grobTree(textGrob("End holidays",
x = 0.756,
y = 0.97,
hjust = 0,
gp = gpar(col = "#393F3E",
fontsize = annotation_size)))
## Plot -------------------------------------------------------------------
for(n_proj_type in c("SQ")){
ggplot(data = subset(df_figS7_PercDiff, proj_type == n_proj_type),
aes(x = dates,
y = value,
colour = Outcome,
ymin = lb,
ymax = ub,
shape = Outcome)) +
geom_point(#position = position_dodge(width = 5),
size = 4) +
geom_errorbar(#position = position_dodge(width = 5),
size = 1.1) +
facet_wrap(~ Outcome, ncol = 2, scales = "free_y") +
# geom_ribbon(aes(x = dates,
# y = mean_percent,
# ymin = lb_percent,
# ymax = ub_percent,
# fill = Intervention),
# alpha = 0.4,
# colour = NA) +
scale_color_manual(values = v_colors_colour) +
scale_fill_manual(values = v_colors_fill) +
scale_shape_manual(values = v_colors_shape) +
guides(shape = guide_legend(title = ""),
color = guide_legend(title = "",
ncol = 1,
override.aes = list(#linetype = v_linetype,
color = v_colors_colour,
# fill = v_colors_fill,
shape = v_colors_shape))) +
geom_vline(xintercept = c(as.Date("2020-12-24"),
as.Date("2021-01-06")),
show.legend = TRUE,
size = 0.5,
linetype = rep(c("dashed", "twodash"),4),
color = "#393F3E") +
# annotation_custom(start_holidays) +
# annotation_custom(end_holidays) +
scale_x_date(label_x_axis,
date_labels = "%b/%d", #Change the format of the date in the x axis
breaks = c(as.Date("2020-12-07"), as.Date("2020-12-14"),
as.Date("2020-12-21"), as.Date("2020-12-28"),
as.Date("2021-01-04"), as.Date("2021-01-11"),
as.Date("2021-01-18"), as.Date("2021-01-25"),
as.Date("2021-02-01"), as.Date("2021-02-08"),
as.Date("2021-02-15"), as.Date("2021-02-22"),
as.Date("2021-03-01"))
# breaks = "week"
) +
scale_y_continuous("Difference from status quo (%)",
breaks = number_ticks(8),
labels = scales::percent
) + # trans = "log"
# labs(# title = "COVID-19 incident cases in MCMA, Mexico",
# caption = gg_caption
# ) +
theme(text = element_text(size = text_size),
plot.caption = element_text(hjust = 0,
size = caption_size),
axis.text.x = element_text(angle = 90,
hjust = 0.5,
vjust = 0.5,
colour = "black"),
axis.text.y = element_text(colour = "black"),
panel.background = element_rect(fill = "white",
colour = "gray",
size = 0.15,
linetype = "solid"),
strip.background = element_rect(fill = "transparent",
colour = "transparent"),
panel.border = element_rect(colour = "black",
fill = NA,
size = 0.7),
legend.position = "none", # c(0.44,0.5),
legend.text = element_text(size = 20),
legend.justification = "center",
legend.direction = "vertical",
legend.title = element_blank(),
legend.key = element_rect(fill = "transparent",
colour = "transparent"))
# Save plot
ggsave(paste0("figs/figs_paper/figS7_PercDiff_SQ_Holidays_",n_proj_type,".jpg"),
width = 21, height = 15, dpi = 300)
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.