COVID_19_HACK: Covid_19 outbreak analysis

#================================
#====== GENERAL DATA ACQUISITION =====
# remotes::install_github("jbkunst/highcharter", upgrade = "never")
countryTS = covid19:::get_countryTS()
regionTS = covid19:::get_regionTS()
provTS = covid19:::get_provTS()
country_growth = covid19:::get_country_growth()
intensivecare_capacity = covid19:::get_intensivecare_cap(regionTS)
age_cases = covid19:::get_agecases(as.character(sort(names(regionTS))))
decrees = covid19:::get_decrees()

N <- nrow(countryTS$Italy)


#===  Global function to check error 
checkExp <- function(expression, message) {
  
  validate(
    
    need( try(expression), message)
  )
  
  return(expression)
  
}

## CHECKS FOR ERROR PREVENTING ##
is_ready <- function(x) {
  if(( is.null(x) || length(x) == 0 ))
    return(FALSE)
  
  return(TRUE)
}

#=== Function to pass from log e to log 10
toLog10<- function(num)
{
  return(log(num)/log(10))
}

# from date to UTS date (milliseconds from Jan 1, 1970)
UTSdate <- function(date) {
  return(as.integer(date) * 86400000)
}


# Aggregates a time series weekly
aggr_wly <- function(v,avg=F) {
  if(avg) {
    out <- unlist( lapply( split(v,ceiling(seq_along(v)/7)) ,mean, na.rm=TRUE) )
  } else {
    out <- unlist( lapply( split(v,ceiling(seq_along(v)/7)) ,sum) )
  }
  return(unname(out))
}

init_date <- min(countryTS$Italy$data)

fin_date <- max(countryTS$Italy$data)

#================================

#================================
#====== MODULE 1 - HOME ====== 

# --- region ---
get_mapregion <- function(category_track){
  
map <- "https://raw.githubusercontent.com/stefanocudini/leaflet-geojson-selector/master/examples/italy-regions.json" %>% 
  httr::GET() %>% 
  httr::content() %>% 
  jsonlite::fromJSON(simplifyVector = FALSE)

dfita1 <-  map$features %>% 
  purrr::map_df(function(x){
    dplyr::as_data_frame(x$properties)
  })

pc_data <- regionTS

names(pc_data) <- tolower(names(pc_data))

pc_df <- purrr::map_df(names(pc_data), function(x){
  dplyr::tibble(
    name=x,
    date=pc_data[[x]]$data,
    cases=pc_data[[x]][[category_track]])
})

pc_df <- pc_df %>%
  dplyr::ungroup() %>% 
  dplyr::group_by(name) %>%
  dplyr::mutate(growth=round(((cases-dplyr::lag(cases))/dplyr::lag(cases))*100,2))

# pc_df$name

# integrate population info
pop_region <- italy_pop$region %>% 
  dplyr::rename(name=territorio,pop=valore) %>%
  dplyr::mutate(name=tolower(name)) %>%
  dplyr::mutate(name=ifelse(name=="emilia romagna", "emilia-romagna", name))

territory_region <- italy_ext$region %>%
  dplyr::rename(name=territorio,ext=valore) %>% 
  dplyr::mutate(name=tolower(name)) %>% 
  dplyr::mutate(name=ifelse(name=="emilia romagna", "emilia-romagna", name))


pc_df <- pc_df %>%
  dplyr::left_join(pop_region) %>% 
  dplyr::left_join(territory_region) %>%
  dplyr::ungroup() %>% 
  dplyr::filter(!name%in%c("friuli v. g. ")) %>%
 dplyr::mutate(name=ifelse(name%in%c("trento","bolzano","p.a. trento","p.a. bolzano"),
                     "trentino-alto adige/sudtirol",name)) %>%
  dplyr::group_by(name, date) %>%
  dplyr::summarise(cases=sum(cases),
            growth=mean(growth,na.rm=T),
            pop=sum(pop),
            ext=sum(ext)) %>%
  dplyr::ungroup() %>% 
 dplyr::mutate(name=ifelse(name=="emilia romagna","emilia-romagna",name))



a <- pc_df$name

b <- dfita1$name

# setdiff(b,a)

dfita1 <- dfita1 %>%
  dplyr::left_join(pc_df) %>%
  dplyr::ungroup() %>% 
 dplyr::mutate(proportion=(cases/pop)*100) %>%
 dplyr::mutate(density=(cases/ext)) %>%
  dplyr::rename(absolute=cases) %>%
 dplyr::mutate(proportion = round(proportion,2)) %>%
 dplyr::mutate(density = round(density, 2)) %>%
  dplyr::select(id, date, absolute, proportion, density, growth) %>%
  dplyr::ungroup() %>%
  tidyr::gather(key="type",value="value",-id,-date)

dfita1

}



# --- province ---

clean_prov <- purrr::map_df(names(provTS), function(x) {
  dplyr::data_frame(
    name=x,
    date=provTS[[x]]$data,
    cases=provTS[[x]]$totale_casi
  )
})

clean_prov <- clean_prov %>%
  dplyr::ungroup() %>% 
  dplyr::group_by(name) %>%
  dplyr::mutate(growth=round(((cases-dplyr::lag(cases))/dplyr::lag(cases))*100,2)) %>%
  dplyr::ungroup()

url <- "http://code.highcharts.com/mapdata/countries/it/it-all.geo.json"
tmpfile <- tempfile(fileext = ".json")

utils::download.file(url, tmpfile)

ita <- readLines(tmpfile)

ita <- gsub(".* = ", "", ita)
ita <- jsonlite::fromJSON(ita, simplifyVector = FALSE)

x <- ita$features[[1]]
x$properties

dfita2 <-  ita$features %>% 
  purrr::map_df(function(x){
    dplyr::data_frame(hasc = x$properties$hasc, name = x$properties$name)
  }) %>%  # extract the keys
  dplyr::mutate(random = runif(nrow(.))) # create random value


# spreading delay

clean_prov_delay <- purrr::map_df(names(provTS), function(x) {
  st_en <- covid19:::detect_start_end(provTS[[x]]$totale_casi)
  dplyr::data_frame(
    name=x,
    start=provTS[[x]][st_en[[1]], "data"],
    end=provTS[[x]][st_en[[2]], "data"],
    peak=provTS[[x]][which.max(diff(provTS[[x]]$totale_casi)), "data"]
  )
}) %>% 
  dplyr::ungroup()

dfita3 <-  ita$features %>% 
  purrr::map_df(function(x){
    dplyr::data_frame(hasc = x$properties$hasc, name = x$properties$name)
  }) %>%  # extract the keys
  dplyr::mutate(random = runif(nrow(.)))




# add population
pop_prov <- dplyr::rename(italy_pop$province, name=territorio,pop=valore)

# add territory
territory_prov <- italy_ext$province %>%
  dplyr::rename(name=territorio,ext=valore)


# make names consistent

clean_prov <- clean_prov %>%
  dplyr::left_join(pop_prov) %>% 
  dplyr::left_join(territory_prov) %>%
  dplyr::ungroup() %>%
  dplyr::mutate(name = ifelse(name=="Massa Carrara","Massa-Carrara",name)) %>%
  dplyr::mutate(name = ifelse(name=="Reggio nell'Emilia","Reggio Emilia",name)) %>% 
  dplyr::mutate(name = ifelse(name=="Bolzano","Bozen",name)) %>%
  dplyr::mutate(name = ifelse(name=="Aosta","Aoste",name)) %>% 
  dplyr::mutate(name = ifelse(name=="Monza e della Brianza","Monza e Brianza",name)) %>%
  dplyr::mutate(name = ifelse(name=="Reggio di Calabria","Reggio Calabria",name)) %>%
  dplyr::mutate(name = ifelse(name=="Torino","Turin",name)) %>%
  dplyr::mutate(name = ifelse(name=="Oristano","Oristrano",name)) %>%
  dplyr::mutate(name = ifelse(name=="Barletta-Andria-Trani","Barletta-Andria Trani",name))

dfita2 <- dfita2 %>%
  dplyr::left_join(clean_prov) %>% 
  dplyr::ungroup() %>% 
  dplyr::mutate(proportion=(cases/pop)*100) %>%
  dplyr::mutate(density=(cases/ext)) %>%
  dplyr::rename(absolute=cases) %>% 
  dplyr::select(hasc, date, absolute, proportion, density, growth) %>%
  tidyr::gather(key="type",value="value",-hasc,-date)


clean_prov_delay <- clean_prov_delay %>%
  dplyr::left_join(pop_prov) %>% 
  dplyr::left_join(territory_prov) %>%
  dplyr::mutate(name = ifelse(name=="Massa Carrara","Massa-Carrara",name)) %>%
  dplyr::mutate(name = ifelse(name=="Reggio nell'Emilia","Reggio Emilia",name)) %>% 
  dplyr::mutate(name = ifelse(name=="Bolzano","Bozen",name)) %>%
  dplyr::mutate(name = ifelse(name=="Aosta","Aoste",name)) %>% 
  dplyr::mutate(name = ifelse(name=="Monza e della Brianza","Monza e Brianza",name)) %>%
  dplyr::mutate(name = ifelse(name=="Reggio di Calabria","Reggio Calabria",name)) %>%
  dplyr::mutate(name = ifelse(name=="Torino","Turin",name)) %>%
  dplyr::mutate(name = ifelse(name=="Oristano","Oristrano",name)) %>%
  dplyr::mutate(name = ifelse(name=="Barletta-Andria-Trani","Barletta-Andria Trani",name))

dfita3 <- dfita3 %>%
  dplyr::left_join(clean_prov_delay) %>% 
  dplyr::ungroup()


# View(dfita3)



#================================


#================================
#====== MODULE 2 - INSPECTION ====== 




# plot growth monitoring --------------------------------------------------------------------
# out_growth <- country_growth
# 
# growth <- data.frame(date=countryTS$Italy$data,
#                      growth=out_growth$growth)
# 
# growth_xts <- xts::xts(growth[,-1], order.by=growth[,1])
# 
# growth_change <- data.frame(date=countryTS$Italy$data,
#                             growth=out_growth$growth_change)
# 
# growth_change_xts <- xts::xts(growth_change[,-1], order.by=growth_change[,1])

# tamponi graph -----------------------------------------------------------

tamp_country <- tibble::tibble(
  data=countryTS$Italy$data,
  tamponi=countryTS$Italy$tamponi,
  totale_casi=countryTS$Italy$totale_casi,
  region = "--- ALL ---"
)

tamp_country_wly <- tibble::tibble(
  data=countryTS$Italy$data[seq(1,N,7)],
  tamponi_settimanali=aggr_wly(diff(c(0,countryTS$Italy$tamponi))),
  casi_settimanali=aggr_wly(diff(c(0,countryTS$Italy$totale_casi))),
  region = "--- ALL ---"
)

tamp_country_avg_wly <- tibble::tibble(
  data=countryTS$Italy$data[seq(1,N,7)],
  tamponi_medi_settimanali=aggr_wly(diff(c(0,countryTS$Italy$tamponi)), avg=T),
  casi_medi_settimanali=aggr_wly(diff(c(0,countryTS$Italy$totale_casi)), avg=T),
  region = "--- ALL ---"
)

tamp_region <- purrr::map_df(names(regionTS), function(x){
  regionTS[[x]] %>%
    dplyr::select(data,tamponi,totale_casi) %>%
    dplyr::mutate(region=x)
})

tamp_region_wly <- purrr::map_df(names(regionTS), function(x){
  data.frame("data"=regionTS[[x]]$data[seq(1,N,7)],
             "tamponi_settimanali"=aggr_wly(diff(c(0,regionTS[[x]]$tamponi))),
             "casi_settimanali"=aggr_wly(diff(c(0,regionTS[[x]]$totale_casi))),
             "region"=x)
})

tamp_region_avg_wly <- purrr::map_df(names(regionTS), function(x){
  data.frame("data"=regionTS[[x]]$data[seq(1,N,7)],
             "tamponi_medi_settimanali"=aggr_wly(diff(c(0,regionTS[[x]]$tamponi)), avg=T),
             "casi_medi_settimanali"=aggr_wly(diff(c(0,regionTS[[x]]$totale_casi)), avg=T),
             "region"=x)
})

tamp_creg <- tamp_country %>% 
  dplyr::bind_rows(tamp_region) %>%
  dplyr::ungroup() %>%
  dplyr::group_by(region) %>%
  dplyr::mutate(casi_giornalieri=totale_casi-dplyr::lag(totale_casi)) %>% 
  dplyr::mutate(casi_giornalieri=ifelse(data==init_date,totale_casi,casi_giornalieri)) %>%
  dplyr::mutate(tamponi_giornalieri=tamponi-dplyr::lag(tamponi)) %>%
  dplyr::mutate(tamponi_giornalieri=ifelse(data==init_date,tamponi,tamponi_giornalieri)) %>%
  dplyr::mutate(share_infected_discovered = casi_giornalieri/tamponi_giornalieri) %>%
  dplyr::select(data,casi_giornalieri,tamponi_giornalieri,share_infected_discovered) %>%
  dplyr::mutate(share_infected_discovered=round(share_infected_discovered,3)) %>%
  dplyr::rename("Daily cases"=casi_giornalieri,"Daily tests"=tamponi_giornalieri,
                "Date"=data)

tamp_creg_1 <- tamp_creg %>% dplyr::select(1:4) %>%
  tidyr::gather(key="key",value="value",-Date, -region)

tamp_creg_wly <- tamp_country_wly %>% 
  dplyr::bind_rows(tamp_region_wly) %>%
  dplyr::ungroup() %>%
  dplyr::group_by(region) %>%
  dplyr::mutate(share_infected_discovered = casi_settimanali/tamponi_settimanali) %>%
  dplyr::select(data,casi_settimanali,tamponi_settimanali,share_infected_discovered) %>%
  dplyr::mutate(share_infected_discovered=round(share_infected_discovered,3)) %>%
  dplyr::rename("Weekly cases"=casi_settimanali,"Weekly tests"=tamponi_settimanali,
                "Date"=data)

tamp_creg_1_wly <- tamp_creg_wly %>% dplyr::select(1:4) %>%
  tidyr::gather(key="key",value="value",-Date, -region)

tamp_creg_avg_wly <- tamp_country_avg_wly %>% 
  dplyr::bind_rows(tamp_region_avg_wly) %>%
  dplyr::ungroup() %>%
  dplyr::group_by(region) %>%
  dplyr::mutate(share_infected_discovered = casi_medi_settimanali/tamponi_medi_settimanali) %>%
  dplyr::select(data,casi_medi_settimanali,tamponi_medi_settimanali,share_infected_discovered) %>%
  dplyr::mutate(share_infected_discovered=round(share_infected_discovered,3),
                casi_medi_settimanali=round(casi_medi_settimanali,3),
                tamponi_medi_settimanali=round(tamponi_medi_settimanali,3)) %>%
  dplyr::rename("Average weekly cases"=casi_medi_settimanali,"Average weekly tests"=tamponi_medi_settimanali,
                "Date"=data)

tamp_creg_1_avg_wly <- tamp_creg_avg_wly %>% dplyr::select(1:4) %>%
  tidyr::gather(key="key",value="value",-Date, -region)

# age_cases ---------------------------------------------------------------

age_df <- purrr::map_df(names(age_cases), function(x) {
  age_cases[[x]]$age_cases
})

age_df <- age_df %>%
  dplyr::filter(!age_int=="Not known") %>%
  dplyr::select(-perc_cases) %>%
  dplyr::ungroup() %>% 
  dplyr::group_by(age_int) %>%
  dplyr::summarise(cases=sum(cases, na.rm=T)) %>%
  dplyr::ungroup() %>%
  dplyr::mutate(perc_cases = round((cases/sum(cases))*100,2)) %>% 
  dplyr::mutate(region="--- ALL ---")

age_df_region <- purrr::map_df(names(age_cases), function(x) {
  age_cases[[x]]$age_cases %>%
    dplyr::mutate(region=x) %>% 
    dplyr::filter(!age_int=="Not known") %>% 
    dplyr::select(-perc_cases) %>%
    dplyr::ungroup() %>% 
    dplyr::mutate(perc_cases = round((cases/sum(cases))*100,2))
})

age_df_final <- age_df %>% 
  dplyr::bind_rows(age_df_region)

#================================




#================================
#====== MODULE 3 - ANALYSIS ====== 

# Inital and final dates of samples
'%then%' <- shiny:::'%OR%'

init_date_arima = init_date

# Total population sizes in 2020 winter
country_tot_pop <- 6.048e+07
region_tot_pop <- NULL

countryNames <- names(countryTS)
regNames <- as.character(sort(names(regionTS)))
provNames <- as.character(sort(names(provTS)))

# Time horizon of all graphs
if(nrow(countryTS$Italy) > 30) {
  days <- c(1:(nrow(countryTS$Italy)+20) )
} else {
  days <- c(1:50)
}

# Association of provinces to regions
regAndProv <- data.frame("province" = provNames, "region" = NA, stringsAsFactors = FALSE)
for(prov in provNames) {
  regAndProv[regAndProv$province == prov, "region"] <- as.character(provTS[[prov]]$denominazione_regione[1])
}

#================================