clean voter file

brookhaven.r

voters <- read.csv("~/Downloads/voter_file.csv", stringsAsFactors = F)
voters_sub <- voters[c("SBOEID","DOB","REGDATE","ENROLLMENT","LASTVOTEDATE","COUNTYCODE","ED","WARD","SD","AD","LD")]
existing_eds <- tbl_df(voters_sub) %>%
    group_by(ED, ENROLLMENT) %>%
    summarise(count = n()) %>%
    spread(ENROLLMENT, count) %>%
    mutate(TOTAL = sum(BLK,CON,DEM,GRE,IND,OTH,REF,REP,WEP,WOR,na.rm=T),
           DEM_PCT = DEM/TOTAL)
new_dems_eds <- tbl_df(voters_sub) %>%
    filter(REGDATE >= 20161109) %>%
    filter(is.na(LASTVOTEDATE)) %>%
    group_by(ED, ENROLLMENT) %>%
    summarise(count = n()) %>%
    spread(ENROLLMENT, count) %>%
    mutate(NEW_TOTAL = sum(BLK,CON,DEM,GRE,IND,OTH,REF,REP,WEP,WOR,na.rm=T),
           NEW_DEM=DEM,
           NEW_DEM_PCT = NEW_DEM/NEW_TOTAL)
eds <- merge(existing_eds[,c("ED","DEM","DEM_PCT","TOTAL")], 
             new_dems_eds[,c("ED","NEW_DEM","NEW_DEM_PCT","NEW_TOTAL")], 
             by="ED")
write.csv(eds, "~/Downloads/brookhaven_eds.csv")