Crimes in Somerville (Time Series)

Handy Packages

library(tidyverse)
library(lubridate)
library(ggthemes)
library(plotly)

Getting the Data

somerville = read.csv("Police_-_Selected_Criminal_Incidents.csv")

Cleaning the Data

• Remove 2018 data since the year has just started.
  
• This part tackles converting the date column into a date data type. Then, I extracted specific parts of the date: the hour, the day of the week, the month, the year, and the day of the month.

somerville$dtreported = mdy_hms(somerville$dtreported)
somerville$wday = wday(somerville$dtreported  , label=TRUE)
somerville$month = month(somerville$dtreported, label=TRUE) #pulls out the month. 
somerville$hour = hour(somerville$dtreported) #pulls out the hour of the day.
somerville$dayomo = mday(somerville$dtreported) #pulls out the day of the month.
somerville$year = year(somerville$dtreported) #pulls out the year. 
somerville = somerville %>%  filter(year !=2018) #removes 2018 data.

Creating the Graph

• I used dplyr to to get a count by hour and day of the week.
  
• I felt fancy so I used ggplotly to make them interactive.

total = somerville %>% count(hour, wday) %>%  #Getting count by hour and day of the week.
  ggplot(aes(wday, hour, fill=n)) + geom_tile() + #create basic heatmap
  scale_fill_gradient2(guide="legend") + theme_tufte() + #legend structure & tufte theme
  theme(text=element_text(size=14), axis.text = element_text(size=14)) + 
  labs(x="Day of the Week", y="Hour", title="Crimes in Somerville (2005-2017)") + #labels
  guides(fill=guide_legend(title="Crimes")) + 
  scale_y_continuous(breaks=seq(0, 23, 3)) #changing y axis

ggplotly(total)

## We recommend that you use the dev version of ggplot2 with `ggplotly()`
## Install it with: `devtools::install_github('hadley/ggplot2')`

It looks like most crimes happen around 9 in the morning on Tuesdays!

But what if we look at the breakdown by type of crime. The data set provides an offense variable which has 4 types of crimes:

• Burglary/breaking and entering
• Motor vehicle theft
• Robbery
  
• Theft from motor vehicle

BURGLARY/BREAKING AND ENTERING

burg = somerville %>% filter(offense=="BURGLARY/BREAKING AND ENTERING") %>% 
  count(hour, wday) %>%  #Getting count by hour and day of the week.
  ggplot(aes(wday, hour, fill=n)) + geom_tile() + #create basic heatmap
  scale_fill_gradient2(guide="legend") + theme_tufte() + #legend structure & tufte theme
  theme(text=element_text(size=14), axis.text = element_text(size=14)) + 
  labs(x="Day of the Week", y="Hour", title="Burglary/Breaking and Entering") + #labels
  guides(fill=guide_legend(title="Crimes")) + 
  scale_y_continuous(breaks=seq(0, 23, 3)) #changing y axis

ggplotly(burg)

It seems that most burglaries happen on Wednesday evenings.

MOTOR VEHICLE THEFT

carTheft = somerville %>% filter(offense=="MOTOR VEHICLE THEFT") %>% 
  count(hour, wday) %>%  #Getting count by hour and day of the week.
  ggplot(aes(wday, hour, fill=n)) + geom_tile() + #create basic heatmap
  scale_fill_gradient2(guide="legend") + theme_tufte() + #legend structure & tufte theme
  theme(text=element_text(size=14), axis.text = element_text(size=14)) + 
  labs(x="Day of the Week", y="Hour", title="Motor Vehicle Theft") + #labels
  guides(fill=guide_legend(title="Crimes")) + 
  scale_y_continuous(breaks=seq(0, 23, 3)) #changing y axis

ggplotly(carTheft)

Car thefts seem to happen on Monday afternoons and around noon time on Thursdays.

ROBBERY

rob = somerville %>% filter(offense=="ROBBERY") %>% 
  count(hour, wday) %>%  #Getting count by hour and day of the week.
  ggplot(aes(wday, hour, fill=n)) + geom_tile() + #create basic heatmap
  scale_fill_gradient2(guide="legend") + theme_tufte() + #legend structure & tufte theme
  theme(text=element_text(size=14), axis.text = element_text(size=14)) + 
  labs(x="Day of the Week", y="Hour", title="Robbery") + #labels
  guides(fill=guide_legend(title="Crimes")) + 
  scale_y_continuous(breaks=seq(0, 23, 3)) #changing y axis

ggplotly(rob)

This makes sense! Most robberies happen Saturday late at night/early in the morning.

THEFT FROM MOTOR VEHICLE

theft = somerville %>% filter(offense=="THEFT FROM MOTOR VEHICLE") %>% 
  count(hour, wday) %>%  #Getting count by hour and day of the week.
  ggplot(aes(wday, hour, fill=n)) + geom_tile() + #create basic heatmap
  scale_fill_gradient2(guide="legend") + theme_tufte() + #legend structure & tufte theme
  theme(text=element_text(size=14), axis.text = element_text(size=14)) + 
  labs(x="Day of the Week", y="Hour", title="Theft from Motor Vehicle") + #labels
  guides(fill=guide_legend(title="Crimes")) + 
  scale_y_continuous(breaks=seq(0, 23, 3)) #changing y axis

#Making it interactive
ggplotly(theft)

It looks like most thefts from motor vehicles happen in the morning on weekdays.

There you have it a quick time series visualiation of crime in Somerville.