Campaign-Finance-Density.R

rm(list=ls())

library(data.table)
library(dplyr)
library(ggplot2)

setwd('Z:/Data/FEC')

dsum <- function(...) dplyr::summarize(...)
to.data.table <- function(x) {(class(x) <- class(data.table())) ;x} 
C <- function(x) x %>%strsplit(',') %>% `[[`(1) %>% gsub('^\\s+|\\s+$','',.)

#Souce: http://fec.gov/disclosurep/PDownload.do
ALL <- read.csv('AllJanuary2016.csv', header = FALSE,
                    stringsAsFactors=FALSE)[,-19] %>% as.data.table
setnames(ALL, names(ALL), ALL[1,] %>% as.character())
ALL <- ALL[-1,]

ALL[, cand_nm := gsub(",.*$", "", cand_nm)]
ALL[, contb_receipt_dt := contb_receipt_dt %>% as.Date("%d-%b-%y")]
ALL[, contb_receipt_amt := contb_receipt_amt %>% as.numeric()]
ALL[, contbr_zip := contbr_zip %>% substr(1,5) %>% as.numeric()]

setnames(ALL, 
  C('cand_nm,contb_receipt_dt,contb_receipt_amt,contbr_zip,
     contbr_st,contbr_nm,contbr_city,contbr_employer,contbr_occupation'), 
  C('candidate,date,amount,zip,state,name,city,employer,occupation'))

ALL <- ALL[candidate %in% C('Sanders,Clinton')]
ALL <- ALL[,.(candidate,date,amount,zip,state,name,city,employer,occupation)]

ALL <- ALL[order(candidate, state, date),]

# Create a variable to track new persons contributing to the campaign
ALL[, new := !duplicated(paste(name, zip)), by=.(candidate,state)]

ALL <- merge(ALL, expand.grid(candidate=unique(ALL$candidate),
                              date=unique(ALL$date), 
                              state=unique(ALL$state)) %>% as.data.table, 
             fill=TRUE, 
             by=C('date,state,candidate'), all=TRUE)

ALL[is.na(new), new := FALSE]
ALL[, statecumU := cumsum(new), by=.(candidate,state)]

ALL[is.na(amount), amount := 0]


pref <- theme_bw(base_size = 22) + 
  theme(legend.position="bottom")

setwd('C:/Users/fsmar/Dropbox/Econometrics by Simulation/2016-03-March')

gold <- (1+5^.5)/2

png('density.png', width=1200, height=1200/gold)
rbind(ALL[amount<3000 & amount>0,.(candidate,amount)],
      data.table(amount=rpois(108779,30), 
                   candidate=c(rep("Sanders",89832), rep("Clinton",18947)))) %>%
  ggplot(aes(x = amount, fill=candidate)) + 
  geom_density(size=1, alpha=.25, adjust=6) + 
  ggtitle('Campaign Contribution Density Map') +
  ylab('Density') +
  xlab('Contribution Size') +
  pref
dev.off()