age.R

#Subset census by those with full-time jobs and are 18 or above.
ratio_age_census = census[(census$full.or.part.time.employment.stat == "Full-time schedules" & 
                             census$age >= 18 & 
                             census$age <= 70),]

#Get percentage of each age with selected income.

#Both
ratio_age = group_by(ratio_age_census, age) %>% 
  summarise(.,
            Both = (sum(instance.weight[X == "50000+." & age >= 18]) / 
                      (sum(instance.weight[X == "50000+." & age >= 18]) + 
                         sum(instance.weight[X == "-50000" & age >= 18]))) * 100)
#Just males
ratio_age_male = group_by(ratio_age_census, age) %>%
  summarise(.,
            Males = (sum(instance.weight[X == "50000+." & sex == "Male" & age >= 18]) / 
                       (sum(instance.weight[X == "50000+." & sex == "Male" & age >= 18]) + 
                          sum(instance.weight[X == "-50000" & sex == "Male" & age >= 18]))) * 100)
#Just females
ratio_age_female = group_by(ratio_age_census, age) %>%
  summarise(.,
            Females = (sum(instance.weight[X == "50000+." & sex == "Female" & age >= 18]) / 
                         (sum(instance.weight[X == "50000+." & sex == "Female" & age >= 18]) + 
                            sum(instance.weight[X == "-50000" & sex == "Female" & age >= 18]))) * 100)

#Inner join all three together.
ratio_age = inner_join(ratio_age, ratio_age_male, by = "age")
ratio_age = inner_join(ratio_age, ratio_age_female, by = "age")

#Stack them for ggplot2.
ratio_age_stacked = with(ratio_age,
                         data.frame(value = c(Both, Males, Females),
                                    variable = factor(rep(c("Both", "Males", "Females"),
                                                          each = NROW(ratio_age))),
                                    age = rep(age, 3)))

#Plot graph.
agePlot = ggplot(ratio_age_stacked, aes(age, value, colour = variable)) + 
  geom_line() +
  theme(legend.title=element_blank()) +
  ggtitle("Income by Age 94' - 95'") + 
  xlab("Age (years)") + 
  ylab("Percent at Age with Full-time Income >50k")
agePlot