New York City Crime & Population Trends

Brent Brewington (github.com/bbrewington) October 2, 2016

Read crime & population data into R:

Scrape NYC Population Data from Wikipedia

nyc.population <- read_html("https://en.wikipedia.org/wiki/Demographics_of_New_York_City") %>% 
     html_nodes(".toccolours") %>% html_table() %>% as.data.frame() %>% 
     .[2:32,1:2] %>% 
     select(year = Historical.population, population = Historical.population.1) %>%     
     mutate(year = as.numeric(year), population = parse_number(population))
glimpse(nyc.population)

## Observations: 31
## Variables: 2
## $ year       <dbl> 1698, 1712, 1723, 1737, 1746, 1756, 1771, 1790, 180...
## $ population <dbl> 4937, 5840, 7248, 10664, 11717, 13046, 21863, 33131...

Get Crime Data from nyc.gov/html/nypd:

1. Downloaded Excel (.xls) file: http://www.nyc.gov/html/nypd/downloads/excel/analysis_and_planning/seven_major_felony_offenses_2000_2015.xls
2. Copy+Paste data into csv file
3. Read into R via readr::read_csv

nyc.crime <- read_csv("nyc_crimedata_2000-2015.csv", col_names = TRUE) %>% 
     gather(year, crime.count, `2000`:`2015`) %>% 
     mutate(year = as.numeric(year))
glimpse(nyc.crime)

## Observations: 112
## Variables: 3
## $ OFFENSE     <chr> "MURDER & NON-NEGL. MANSLAUGHTER", "RAPE", "ROBBER...
## $ year        <dbl> 2000, 2000, 2000, 2000, 2000, 2000, 2000, 2001, 20...
## $ crime.count <int> 673, 2068, 32562, 25924, 38352, 49631, 35442, 649,...

Plot of NYC population trend

Fit linear model to NYC population (1980-2010)

## 
## Call:
## lm(formula = nyc.population1$population ~ nyc.population1$year)
## 
## Residuals:
##       1       2       3       4       5 
##   40845 -124691  144572 -105044   44317 
## 
## Coefficients:
##                       Estimate Std. Error t value Pr(>|t|)   
## (Intercept)          -75428524    9289882  -8.119  0.00391 **
## nyc.population1$year     41646       4648   8.961  0.00293 **
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 130500 on 3 degrees of freedom
## Multiple R-squared:  0.964,  Adjusted R-squared:  0.952 
## F-statistic: 80.29 on 1 and 3 DF,  p-value: 0.002933

NYC population from 1980-2010 = -75428524 + (41646 * year)

Create dataset: NYC Crime Rate 2000-2015

nyc.population_crime <- data.frame(year = 2000:2015) %>% 
     mutate(population.est = -75428524 + 41646 * year) %>%
     left_join(nyc.crime, by = "year") %>%
     mutate(crime.rate = crime.count / population.est) %>% 
     mutate(crime.rate_per.million = crime.rate * 1000000)

## Observations: 112
## Variables: 6
## $ year                   <dbl> 2000, 2000, 2000, 2000, 2000, 2000, 200...
## $ population.est         <dbl> 7863476, 7863476, 7863476, 7863476, 786...
## $ OFFENSE                <chr> "MURDER & NON-NEGL. MANSLAUGHTER", "RAP...
## $ crime.count            <int> 673, 2068, 32562, 25924, 38352, 49631, ...
## $ crime.rate             <dbl> 8.558556e-05, 2.629880e-04, 4.140917e-0...
## $ crime.rate_per.million <dbl> 85.58556, 262.98802, 4140.91682, 3296.7...

Create dataset: NYC Violent Crime Rate 2000-2015

nyc.population_violentcrime <- data.frame(year = 2000:2015) %>% 
     mutate(population.est = -75428524 + 41646 * year) %>%
     left_join(nyc.crime %>% filter(OFFENSE %in% c("MURDER & NON-NEGL. MANSLAUGHTER", "RAPE", "ROBBERY", "FELONY ASSAULT")) %>% group_by(year) %>% 
                    summarise(crime.count = sum(crime.count)), by = "year") %>%
     mutate(crime.rate = crime.count / population.est, 
            crime.rate_per.million = 1000000 * crime.count / population.est)

## Observations: 16
## Variables: 5
## $ year                   <dbl> 2000, 2001, 2002, 2003, 2004, 2005, 200...
## $ population.est         <dbl> 7863476, 7905122, 7946768, 7988414, 803...
## $ crime.count            <int> 61227, 54285, 51107, 47795, 45470, 4486...
## $ crime.rate             <dbl> 0.007786251, 0.006867067, 0.006431168, ...
## $ crime.rate_per.million <dbl> 7786.251, 6867.067, 6431.168, 5983.040,...

Plots

ggplot(nyc.population_crime, aes(year, crime.rate_per.million, color = OFFENSE)) + geom_point() + geom_smooth(se = FALSE) + ggtitle("NYC Seven Major Felony Offenses by Year: 2000-2015") +
     xlab("Year") + ylab("Seven Major Felonies Per Million NYC Residents")

ggplot(nyc.population_crime %>% filter(OFFENSE == "MURDER & NON-NEGL. MANSLAUGHTER"), aes(year, crime.rate_per.million)) + geom_point() + geom_smooth() + geom_line() +
     xlab("Year") + ylab("Murders Per Million NYC Residents") + ggtitle("NYC Murder Rate: 2000-2015") + geom_label(aes(x = 2010, y = 90), label = "Blue Line: Lowess Smoothing (with 95% confidence band)") + scale_y_continuous(limits = c(0, 100))

ggplot(nyc.population_violentcrime, aes(year, crime.rate_per.million)) + geom_bar(stat = "identity", aes(fill = "Total Violent Crime Rate")) + geom_smooth(data = nyc.population_crime %>% filter(OFFENSE %in% c("MURDER & NON-NEGL. MANSLAUGHTER", "RAPE", "ROBBERY", "FELONY ASSAULT")), aes(x = year, y = crime.rate_per.million, color = OFFENSE), se = FALSE) + geom_point(data = nyc.population_crime %>% filter(OFFENSE %in% c("MURDER & NON-NEGL. MANSLAUGHTER", "RAPE", "ROBBERY", "FELONY ASSAULT")), aes(x = year, y = crime.rate_per.million, color = OFFENSE)) + scale_fill_manual(values = "black", name = NULL) + scale_color_discrete(name = NULL) +
     ggtitle("NYC Violent Crime Rate: 2000-2015") + ylab("Violent Crimes per Million NYC Residents") + xlab("Year")