Choropleth - 2013 Crime in America Data

Exploring 2013 crime data and building a choropleth map to represent violent crime by state

1. Read in data removing footnotes

crime <- read.csv("crime.csv")  
crime <- crime[-c(9293:9302),]

2. Examine data structure

The data is arranged as one giant merged cell for each state with many rows representing cities within it

str(crime)  
which(crime$State != "")

3. Fill in missing state names and clean up

library(zoo)  
States <- (crime$State)  
States[States == ""] <- NA  
States <- na.locf(States, na.rm = TRUE, fromLast = FALSE)  
crime$States <- States  
crime <- crime[,-c(1)]  
crime <- crime[,-c(14:16)]  

crime[is.na(crime)] <- 0  

4. Aggregate data by state

For now we're more interested in crime at the state level rather than city

crimeAgg <- aggregate(. ~ States, data = crime, sum)  

5. Calculate percent crime per capita for variables of interest

crimeAgg$ViolentPerCap <- (crimeAgg$Violent.crime / crimeAgg$Population) * 100
crimeAgg$BurglaryPerCap <- (crimeAgg$Burglary / crimeAgg$Population) * 100
crimeAgg$VehiclePerCap <- (crimeAgg$Motor.vehicle.theft / crimeAgg$Population) * 100

Preliminarily, we can examine crime by state in ggplot2

library(ggplot2)  
ggplot(crimeAgg, aes(x=States, y=VehiclePerCap)) + geom_col()  
ggplot(crimeAgg, aes(x=States, y=ViolentPerCap)) + geom_col()  
ggplot(crimeAgg, aes(x=States, y=BurglaryPerCap)) + geom_col()  

5. Plot choropleth map of violent crime by state

library(choroplethr)  
library(choroplethrMaps)  
violentCrime <- crimeAgg[,-c(2:14,16, 17, 18)]  
colnames(violentCrime) <- c("region", "value")  
violentCrime$region <- sapply(violentCrime$region, tolower)  
violentCrime$value <- as.numeric(as.character(violentCrime$value))  
state_choropleth(violentCrime, title = "Violent Crime, 2013", legend = 
                 "Percent violent crime per capita", num_colors = 1, reference_map = FALSE)