## Creates single file containing median variables at the
## tract level for DC Area from the 2010 Census and 2010-2014 through
## 2013-2017 American Community Surveys for tracts in the DC Area

## Set up environment
library(tidyverse)

agevar <- 'mdage'

## Identify variables to keep that are not race variables
geovars <- c('GISJOIN', 'STATE', 'COUNTY')

## Set Functions
source('dcarea_functions.R')

## Create dataset that contains only DC-area tracts and constructed
## race variables
select.agevars <- function(dta) {
    dta <- select.dcarea(dta)
    dta$GISJOIN <- as.character(dta$GISJOIN)
    return(dta[, c(geovars, agevar)])
}

## Load datasets containing median age variables from 2010 TIGER/Line files
load('../US/tracts/2010/tabular/median-age/dataset/tracts-2010TIGER-median-age.Rdata')

## Create list of data frames for each year and restrict to created race
## variables
dtas <- list(trt10c_ag, trt14_ag, trt15_ag, trt16_ag, trt17_ag)
agedtas <- lapply(dtas, select.agevars)

## Create single wide data frame with year appended to variable name
suffixes <- c('10c', 14:17)
namelist <- lapply(suffixes, function(x) c(geovars, paste0(agevar, x)))
agedta <- mapply(setNames, agedtas, namelist, SIMPLIFY = FALSE) %>%
    reduce(left_join, by='GISJOIN')
agedta$STATE <- agedta$STATE.x
agedta$COUNTY <- agedta$COUNTY.x
agedta <- agedta[, -(grep('^STATE.+|^COUNTY.+', names(agedta), perl=TRUE))]

## Write file containing wide dataset to CSV format
write.csv(agedta, 'tracts/2010/tabular/median-age/dataset/tracts-2010TIGER-median-age.csv')