Family groups and income deciles from NZ HES

HES_family.R

HES9 <- read.csv("household-income-and-houseing-statistics-ye-june2-17-csv-tables-corrected/HES2017-Table9.csv",
         stringsAsFactors = FALSE)

library(dplyr)
library(tidyr)
library(purrr)
library(ggplot2)
library(ggbeeswarm)
library(ggthemes)
atomise <- function(df) {
  df2 <- data.frame(disagg = rep(df$d2, round(df$households, 0)))
  df2$adj <- round((1:nrow(df2))/nrow(df2),1) *.94 - 0.47
  df2$decil <- df2$disagg + df2$adj
  return(df2)
}
one_parent <- HES9 %>% filter(Rcode == "HHC06") %>%
  select(Year, Est1:Est10) %>%
  gather(key=dectext, value=households, Est1:Est10, convert=TRUE) %>%
  mutate(decile= as.numeric(gsub("Est", "", dectext)),
         d2 = decile) %>% 
  filter(!is.na(households)) %>%
  select(-dectext) %>% 
  group_by(Year, decile) %>%
  nest() %>%
  mutate(dec = map(data, atomise)) %>%
  select(-data) %>%
  unnest(dec) %>%
  ungroup() %>%
  mutate(household_type ="One parent households")

HES9 %>% filter(Rcode %in% c("HHC02", "HHC03", "HHC04", "HHC05")) %>%
  select(Year, Est1:Est10) %>%
  gather(key=dectext, value=households, Est1:Est10, convert=TRUE) %>%
  filter(!is.na(households)) %>% 
   group_by(Year, dectext) %>%
  summarise(households = sum(households)) %>%
  ungroup() %>%
   mutate(decile= as.numeric(gsub("Est", "", dectext)),
         d2 = decile) %>% 
  select(-dectext) %>% 
  group_by(Year, decile) %>%
  nest() %>%
  mutate(dec = map(data, atomise)) %>%
  select(-data) %>%
  unnest(dec) %>%
  ungroup() %>%
  mutate(household_type ="Two parent households") %>%
  bind_rows(one_parent) %>%
  ggplot(aes(x=Year, y=decil, colour=as.factor(decile))) + 
  geom_quasirandom(width=0.41, size=0.2) +
  theme_tufte() + facet_wrap(~ household_type, nrow=2) +
  geom_hline(yintercept = 0.5, colour="#EEEEEE") +
  geom_hline(yintercept = 1.5, colour="#EEEEEE") +
  geom_hline(yintercept = 2.5, colour="#EEEEEE") +
  geom_hline(yintercept = 3.5, colour="#EEEEEE") +
  geom_hline(yintercept = 4.5, colour="#EEEEEE") +
  geom_hline(yintercept = 5.5, colour="#EEEEEE") +
  geom_hline(yintercept = 6.5, colour="#EEEEEE") +
  geom_hline(yintercept = 7.5, colour="#EEEEEE") +
  geom_hline(yintercept = 8.5, colour="#EEEEEE") +
  geom_hline(yintercept = 9.5, colour="#EEEEEE") +
  geom_hline(yintercept = 10.5, colour="#EEEEEE") +
  ylab("decile") + 
  ggtitle("HES income deciles by family type,\n1 dot=1000 households") +
  scale_y_continuous(breaks=1:10) +
  theme(legend.position="none")