top 1 percent ipums

top_1_percent

# File # 1
# https://drive.google.com/open?id=1Lhz23JP4gRW4p_01D7OzaL60CT1wfem-

# File # 2
# https://drive.google.com/open?id=1XOWD8COTdDx30HNLZgrp57WzWUGOtaKA

# NOTE: To load data, you must download both the extract's data and the DDI
# and also set the working directory to the folder with these files (or change the path below).

#install.packages("tidyverse")
library(tidyverse)

install.packages("ipumsr")
library(ipumsr)

# Put the two files (usa_00008.dat.gz and usa_00008.xml) into your working directory

ddi <- read_ipums_ddi("usa_00008.xml")
data <- read_ipums_micro(ddi)

install.packages("Hmisc")
library(Hmisc)

View(data)

data %>%
  select_if(is.labelled)

ipums_val_labels(data$STATEFIP)

# Convert the labels to factors (and drop the unused levels)
data <- data %>%
  mutate(STATE_factor = as_factor(lbl_clean(STATEFIP)),
         COUNTY_factor = as_factor(lbl_clean(COUNTY)),
         COUNTYFIPS_factor = as_factor(lbl_clean(GQ)),
         HHINCOME_factor = as_factor(lbl_clean(HHINCOME)))



md <- filter(data, STATE_factor=="Maryland")

View(md)

mont <- filter(data, COUNTYFIPS==31)


View(mont)

# wrong


wtd.quantile(mont$HHINCOME, mont$HHWT, probs=c(.99))

range(mont$HHINCOME)

# let's fix

mont2 <- mont %>%  mutate(
  HHINCOME = as.numeric(as.character(HHINCOME)),
  HHINCOME=case_when(
    HHINCOME==9999999 ~ 0,
    TRUE ~ HHINCOME
))

# What's the range now?

range(mont2$HHINCOME)

wtd.quantile(mont2$HHINCOME, mont2$HHWT, probs=c(.99))

# Now let's see where the county ranks in the country

ranked <- filter(data, HHINCOME!=9999999) %>%
  group_by(STATE_factor) %>%
  summarize(one_percent=wtd.quantile(HHINCOME, HHWT, probs=c(.99))) %>%
  mutate(rank=rank(desc(one_percent)))

View(ranked)

# Prep the data

data <- mutate(data,
               statef=as.character(STATEFIP),
               countyf=as.character(COUNTYFIPS),
               statef=case_when(
                 nchar(statef)==1 ~ paste0("0", statef),
                 TRUE ~ statef),
               countyf=case_when(
                 nchar(countyf)==1 ~ paste0("00", countyf),
                 nchar(countyf)==2 ~ paste0("0", countyf),
                 TRUE ~ countyf),
               fips=paste0(statef, countyf))

# Rank by all available counties

# PERNUM is persons per household.
# We want this to be 1 so it's not double counting
# https://usa.ipums.org/usa-action/variables/PERNUM#description_section [usa.ipums.org]
# and GQ stands for Group Housing status
# https://usa.ipums.org/usa-action/variables/PERNUM#description_section [usa.ipums.org]
# Everything higher than 3 is institutions and dorms or group homes, etc

ranked <- filter(data, HHINCOME!=9999999) %>%
  filter(PERNUM==1) %>%
  filter(GQ<3) %>%
  group_by(STATE_factor, fips) %>%
  summarize(one_percent=wtd.quantile(HHINCOME, HHWT, probs=c(.99)),
            
            count=n()) %>%
  ungroup() %>% 
  mutate(rank=rank(-one_percent))