riseR:

# OBJECTIVE 3 ANNUAL SURVEY - INDONESIA: HOUSEHOLD - 11 Nov 2019 to ??

library (tidyverse)
library (lubridate)
library (stringr)

rm(list = ls())
setwd("S:/R-MNHS-SPHPM-EPM-IDEpi/RISE/4. Surveys/3. Objectives/2. ID/3/20191118_T4_annual_survey/2. Data/1. raw data")


# DEFINE TODAY FOR DAILY QC - this is the date of data collection
date <- "2019-11-11"
day.qc <- ymd (date)
rm(date)

###############
# OPEN DATA FILES
###############

# HOUSEHOLD SURVEY
hhd <- read_csv (file = "T4_ID_annual_v1.csv")
child_loop <- read_csv (file = "T4_ID_annual_v1-child_loop.csv")
child_feces_kit <- read_csv (file = "T4_ID_annual_v1-feces_kit.csv")

#upload all files so corrections script can run 
#add this once collecting feces samples
setwd("S:/R-MNHS-SPHPM-EPM-IDEpi/RISE/4. Surveys/3. Objectives/2. ID/3/20190527_O3_4_6mo_survey/3. ID/2. Data/1. raw data/practice data")
feces <- read_csv (file = "RISE_O3_feces_ID_v2_practice.csv")
child_feces_collect <- read_csv (file = "RISE_O3_feces_ID_v2_practice-feces_collection.csv")

###############
# FIX DATES
###############
# need to fix dates to allow for return visits with survey left open (can't use "today")
hhd$endtime <- date(dmy_hms (hhd$endtime)) 

#############################################
##  Correct known errors in the data       ##
#############################################
# setwd("S:/R-MNHS-SPHPM-EPM-IDEpi/RISE/4. Surveys/3. Objectives/2. ID/3/20190902_O3_stool sampling/2. Data/2. code")
# source("O3_T3_ID-corrections.R")
# setwd("S:/R-MNHS-SPHPM-EPM-IDEpi/RISE/4. Surveys/3. Objectives/2. ID/3/20190902_O3_stool sampling/2. Data")


#MERGE
child <- full_join(hhd, child_feces_kit, by = c("KEY" = "PARENT_KEY"))
child_survey <- full_join(hhd, child_loop, by = c("KEY" = "PARENT_KEY"))

feces_pickup <- full_join(feces, child_feces_collect, by = c("KEY" = "PARENT_KEY"))


#############################################
#############################################
# SUBSET - ONLY DATA COLLECTED TODAY
#############################################
#############################################

subhhd <- subset (hhd, endtime == day.qc, 
                  select = c (extract_settlement, settlement_barcode, extract_house_no, gift_yn, feces_participate, verify1, 
                              duration15, time12, time13))
subchild <- subset (child, endtime == day.qc, 
                    select = c (extract_settlement, settlement_barcode, extract_house_no, feces_kit_barcode, 
                                feces_kit_barcode_note))
subchildsurvey <- subset (child_survey, endtime == day.qc, 
                    select = c (extract_settlement, settlement_barcode, extract_house_no, child_no_survey, health_general_child))
subfeces <- subset (feces_pickup, endtime == day.qc, 
                    select = c (extract_settlement, settlement_barcode, extract_house_no, feces_sample_yn, barcode_feces, 
                                barcode_feces_text))

#############################################
#############################################
# DAILY QC REPORT
#############################################
#############################################

#############################################
####### what settlements were visited #######
#############################################

#household surveys:
settlement_hhd <-  subhhd %>%
  select (settlement_barcode)

# then group
settlement.hhd <- settlement_hhd %>% 
  arrange (settlement_barcode) %>% 
  group_by (settlement_barcode) %>%
  summarize (count = n ())

settlement.list.hhd <- pull (settlement.hhd, var = settlement_barcode)

#feces pickup
settlement_feces <-  subfeces %>%
  select (settlement_barcode)

# then group
settlement.feces <- settlement_feces %>% 
  arrange (settlement_barcode) %>% 
  group_by (settlement_barcode) %>%
  summarize (count = n ())

settlement.list.feces <- pull (settlement.feces, var = settlement_barcode)

#############################################
# HOUSEHOLD SURVEYS
#############################################

#number started
nrow(subhhd)

#number of hhd surveys
hhd.complete <- subhhd %>% 
  filter (!is.na(gift_yn))

#############################################
# FECES KITS HANDED OUT
#############################################

#CHILD
kit_child <- subchild %>% 
  filter (!is.na(feces_kit_barcode) | !is.na(feces_kit_barcode_note))

#############################################
# health surveys completed
#############################################
#CHILD
survey_child <- subchildsurvey %>% 
  filter (is.na(child_no_survey) & !is.na(health_general_child))


#############################################
# feces samples collected
#############################################

#CHILD
feces_samples <- subfeces %>% 
  filter (!is.na(barcode_feces) | !is.na(barcode_feces_text))

#############################################
# survey length
#############################################
#FULL SURVEY
survey_time <- subhhd %>% 
  filter (gift_yn==1) %>%  #only surveys that progressed; someone at home
  mutate (dur15 = duration15/60)
summary(survey_time$dur15)

#collective efficacy and action
o5 <- subhhd %>% 
  filter (gift_yn==1) %>%  #only surveys that progressed; someone at home
  mutate (t12 = ymd_hms (time12), 
          t13 = ymd_hms (time13), 
          dur = (t13 - t12)/60)
summary(as.integer(o5$dur))





#############################################
# VERIFICATION ERRORS
# check that all surveys collected were verified by respondents
hhd.no.verify <- subhhd %>% 
  filter (verify1 == 0)


feces.no.verify <- subfeces %>% 
  filter (verify1 == 0)