houses.r

#load libraries
library(readr)
library(dplyr)
library(ggplot2)
library(RColorBrewer)
library(scales)

#load data
ipums <- read_csv('data/usa_00010.csv', col_types=cols(HHWT=col_double(),PERWT=col_double()))

#Remove Alaska and Hawaii
a <- ipums %>% filter(!(YEAR < 1960 & STATEFIP %in% c(2,15)))

#Make factors for region and sex
b <- a %>% mutate(REGION=factor(ifelse(STATEFIP %in% c(17, 18, 19, 20, 26, 27, 29, 31, 38, 39, 46, 55), 1,
                  ifelse(STATEFIP %in% c(4, 6, 8, 16, 30, 32, 35, 41, 49, 53, 56), 2,
                  ifelse(STATEFIP %in% c(9, 23, 25, 33, 34, 36, 42, 44, 50), 3, 4))),
                  labels=c('Midwest','West','Northeast','South')))

c <- b %>% mutate(SEX==1, 'male', 'female')

#Remove group quarters
d<- c %>% filter(GQ==1)

#Create data set for head of house
e <- d %>% filter(RELATE==1)
f <- e %>% mutate(HRace=RACE)

#Create data set for spouses of head of house
g <- d %>% filter(RELATE==2)
h <- g %>% mutate(SRace=RACE)
i <- h %>% select(YEAR, SERIAL,PERNUM, SRace)

#Merge data sets to make coupling
j <- left_join(f,i, by=c('YEAR','SERIAL','SPLOC'='PERNUM'))

#Group into 
k <- j %>% mutate(Status=factor(ifelse(SPLOC==0,1,
                          ifelse(HRace!=SRace, 2, 3)),
                          labels=c('Head Unmarried', 'Different Races', 'Same Race')))

#Total up by region and year to get total households
l <- k %>% group_by(YEAR,REGION) %>% summarize(Number=sum(HHWT))

#Group by type of household
m <- k%>% group_by(YEAR,REGION,Status) %>% summarize(Total=sum(HHWT))

#Join them so they can be divided
n <- left_join(m,l,by=c('YEAR','REGION'))

#Divide to get percent of each type
o <- n %>% mutate(Percent=Total/Number)

#Create Spine Graph
graph1 <- ggplot(o,aes(x=REGION, y=Percent, fill=Status)) +
  geom_bar(aes(width=rescale(Number, c(.1,1))), stat='identity') +
  scale_y_continuous(labels=scales::percent) +
  scale_fill_brewer(palette='Set2') +
  facet_wrap(~YEAR, ncol = 5)+
  labs(title = 'Percent of Households Headed by Couples of Same Race and Different Races, 1900-1990', x = 'Region', y = 'Percent of Household', fill = 'Head and Spouse')

graph2 <- graph1 +
  geom_text(label=ifelse(o$Status=='Different Races',paste('Different ',round(o$Percent*100,1),'%',sep=''),''),
            y=ifelse(o$Status=='Different Races',.5,.9),angle=90)
ggsave('PercentCouples.pdf',width=20, height=7.5)