Setup

Get the socviz package and tidyverse. Build the counties dataframe as in the previous set of exercises.

library(socviz)
library(tidyverse)

## ── Attaching packages ─────────────────────────────────────── tidyverse 1.3.1 ──

## ✓ ggplot2 3.3.3     ✓ purrr   0.3.4
## ✓ tibble  3.1.2     ✓ dplyr   1.0.6
## ✓ tidyr   1.1.3     ✓ stringr 1.4.0
## ✓ readr   1.4.0     ✓ forcats 0.5.1

## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## x dplyr::filter() masks stats::filter()
## x dplyr::lag()    masks stats::lag()

counties = county_data %>% 
  mutate(numeric_fips = as.numeric(id)) %>% 
  filter(numeric_fips %% 1000 != 0 &
         numeric_fips %/% 1000 != 11) %>% 
  mutate(state = factor(state)) %>% 
  filter( pop > 0 )

Trim

Let’s reduce the number of variables in counties and add a continuous variable density.

Answer

counties = counties %>% 
  mutate(density = pop/land_area) %>%
  select(fips,name,census_region,state,pop,land_area, density)

Make Long and Narrow

Convert counties to counties_long by collapsing pop, density, and land_area into a variable measure.

Answer

counties_long = counties %>% 
  pivot_longer(cols = pop:density, names_to = "measure", values_to = "value")

head(counties_long)

## # A tibble: 6 x 6
##    fips name           census_region state measure      value
##   <dbl> <chr>          <fct>         <fct> <chr>        <dbl>
## 1  1001 Autauga County South         AL    pop        55395  
## 2  1001 Autauga County South         AL    land_area    594. 
## 3  1001 Autauga County South         AL    density       93.2
## 4  1003 Baldwin County South         AL    pop       200111  
## 5  1003 Baldwin County South         AL    land_area   1590. 
## 6  1003 Baldwin County South         AL    density      126.

Graph

Use facet_grid() to show how the regions compare for these measures.

Answer

counties_long %>% 
  ggplot(aes(x = value)) +
  geom_density() + 
  geom_rug() + 
  scale_x_log10() +
  facet_grid(census_region ~ measure, scales = "free_x") 

Smaller

That is a bit busy and possibly hard to read. Use filter to compare just the West and the Northeast for land area.

Answer

counties_long %>% 
  filter(census_region %in% c("West","Northeast") &
           measure == "land_area") %>% 
  ggplot(aes(x = value)) +
  geom_density() + 
  geom_rug() + 
  scale_x_log10() +
  facet_grid(census_region ~ measure) 

Summarize

Let’s use summarize to compare the land areas of counties in the regions. Capture min, median, and max values.

Answer

counties_long %>% 
  filter(measure == "land_area") %>% 
  group_by(census_region) %>% 
  summarize(min = min(value),
            median = median(value),
            max = max(value)) %>% 
  ungroup() %>% 
  arrange(median) %>% 
  ggplot(aes(y = census_region)) +
     geom_point(aes(x = min),color = "blue") +
     geom_point(aes(x = median),color = "black") +
     geom_point(aes(x = max),color = "red") +
  ggtitle("Comparison of Land Areas")

Fix the Compression

Use a log scale on the x-axis

Answer

counties_long %>% 
  filter(measure == "land_area") %>% 
  group_by(census_region) %>% 
  summarize(min = min(value),
            median = median(value),
            max = max(value)) %>% 
  ungroup() %>% 
  arrange(median) %>% 
  ggplot(aes(y = census_region)) +
     geom_point(aes(x = min),color = "blue") +
     geom_point(aes(x = median),color = "black") +
     geom_point(aes(x = max),color = "red") +
  ggtitle("Comparison of Land Areas") +
  scale_x_log10()

Joining the Uninsured Rate

The following code will create a dataframe containing the uninsured rate by county.

URL <- "https://healthinequality.org/dl/health_ineq_online_table_12.csv"
characteristicsData <- read.csv(URL)
uninsured_rate = characteristicsData %>% 
  select(cty,puninsured2010,county_name,statename)
str(uninsured_rate)

## 'data.frame':    3138 obs. of  4 variables:
##  $ cty           : int  1001 1003 1005 1007 1009 1011 1013 1015 1017 1019 ...
##  $ puninsured2010: num  13.6 19.1 18.5 17.7 19.3 ...
##  $ county_name   : chr  "Autauga" "Baldwin" "Barbour" "Bibb" ...
##  $ statename     : chr  "Alabama" "Alabama" "Alabama" "Alabama" ...

Now I want to see the distribution of county uninsured rates by census_region.

Get Regions Table

Trim the counties dataframe down to just the identifying variables.

Answer

regions = counties %>% 
  select(fips,census_region, state, name)
head(regions)

##   fips census_region state           name
## 1 1001         South    AL Autauga County
## 2 1003         South    AL Baldwin County
## 3 1005         South    AL Barbour County
## 4 1007         South    AL    Bibb County
## 5 1009         South    AL  Blount County
## 6 1011         South    AL Bullock County

joined = regions %>% 
  left_join(uninsured_rate,by = c("fips" = "cty"))

head(joined)

##   fips census_region state           name puninsured2010 county_name statename
## 1 1001         South    AL Autauga County       13.60128     Autauga   Alabama
## 2 1003         South    AL Baldwin County       19.08533     Baldwin   Alabama
## 3 1005         South    AL Barbour County       18.51381     Barbour   Alabama
## 4 1007         South    AL    Bibb County       17.71883        Bibb   Alabama
## 5 1009         South    AL  Blount County       19.28041      Blount   Alabama
## 6 1011         South    AL Bullock County       16.24321     Bullock   Alabama

Graph

Use facet_wrap() to see the distribution of the uninsured rate by region.

Answer

joined %>% 
  ggplot(aes(x = puninsured2010)) +
  geom_density() +
  geom_rug() +
  facet_wrap(~census_region,ncol = 1)

## Warning: Removed 10 rows containing non-finite values (stat_density).