visuals

# append lists by domain for visuals
setwd("D:/Dropbox/Dropbox (Dimagi)/R analysis/tables/visual_monthly")
file_dir <- ("D:/Dropbox/Dropbox (Dimagi)/R analysis/tables/monthly_table_temp")
filelist <- dir(file_dir, pattern = ".csv", recursive = TRUE, all.files = TRUE, full.names = TRUE)
mt <- lapply(filelist, read.csv)
mt_merged <- do.call("rbind", mt)
mt_merged <- mt_merged[, c(4, 3, 10, 34, 35, 5:9, 11:33, 36:38)]
mt_per_dm <- split(mt_merged, mt_merged$domain.index) # this returns a list of dataframes for each domain that has flw with at least one full calendar month on their CC lifetime
	# or use data.table() when levels increase since split would be slow
setwd("D:/Dropbox/Dropbox (Dimagi)/R analysis/tables/monthly_table_per_domain")
monthlyPerDomainOut <- function(x) {
	for (i in seq_along(x)) {
		filename = paste("domain_", x[[i]]$domain.index[1], sep = "", ".csv")
		write.csv(x[[i]], filename)
	}
}
monthlyPerDomainOut(mt_per_dm)

## work off list mt_per_dm & produce monthly trend of each flw in the same domain
	## guide_legend to be hidden (overflowing for mvp domains since they have huge number of users)
library(ggplot2)

viz_per_dm <- lapply(mt_per_dm, function(x) {
	x$month.index <- as.factor(as.yearmon(x$month.index))
	for (i in 6:(length(x)-3)) { # let's hold on for the viz of outcome indicators for now
		p <- ggplot(x[!is.na(colnames(x)[i]),], aes_string(x = colnames(x)[2], y = colnames(x)[i], color = colnames(x)[3])) +
			geom_point(aes_string(size = colnames(x)[i])) + geom_line(aes_string(group = colnames(x)[3])) + theme(axis.text.x = element_text(angle = 90, hjust = 1), legend.title=element_blank())
		print(p)
		ggsave(filename = paste("dm_", x$domain.index[1], "_", colnames(x)[i], sep = "", ".pdf"), plot = p)	
	}
})