San Francisco incident rates Summer 2014
Giulia Gualdi
9 giugno 2016
setwd('/Users/giulia/datasci_course_materials/Assignments/assignment6')
library(ggplot2)#data visualization
library(ggthemes) #data visualization
library(plyr) #data manipulation
library(dplyr) #data manipulation
library(lubridate) #date manipulation
library(RColorBrewer) #data visualization
library(maptools)#map manipulation
library(rgeos) #map manipulation
library(scales) #data visualization
library(ggmap)#map visualizationKey take away
Thieves are particularly active in San FranCisco and like the Eastern coast and the afternoon hours better. If you’re in the Southern district though, you should keep your eyes open all the day!
Load data
SanFran =read.csv('sanfrancisco_incidents_summer_2014.csv',na.strings=c(""))
#Seattle = read.csv('seattle_incidents_summer_2014.csv',na.strings=c(""))San Francisco
Exploring the dataset
First, I give a look at the data
summary(SanFran)## IncidntNum Category
## Min. : 10284385 LARCENY/THEFT :9466
## 1st Qu.:140545607 OTHER OFFENSES:3567
## Median :140632022 NON-CRIMINAL :3023
## Mean :142017280 ASSAULT :2882
## 3rd Qu.:140719664 VEHICLE THEFT :1966
## Max. :990367398 WARRANTS :1782
## (Other) :6307
## Descript DayOfWeek Date
## GRAND THEFT FROM LOCKED AUTO: 3766 Friday :4451 06/28/2014: 410
## STOLEN AUTOMOBILE : 1350 Monday :4005 08/09/2014: 410
## LOST PROPERTY : 1202 Saturday :4319 08/08/2014: 403
## PETTY THEFT OF PROPERTY : 1125 Sunday :4218 06/29/2014: 397
## WARRANT ARREST : 980 Thursday :3968 08/29/2014: 388
## PETTY THEFT FROM LOCKED AUTO: 955 Tuesday :3930 06/04/2014: 380
## (Other) :19615 Wednesday:4102 (Other) :26605
## Time PdDistrict Resolution
## 12:00 : 784 SOUTHERN :5739 NONE :19139
## 00:01 : 661 MISSION :3700 ARREST, BOOKED : 6502
## 18:00 : 649 NORTHERN :3589 ARREST, CITED : 1419
## 19:00 : 621 CENTRAL :3513 LOCATED : 1042
## 17:00 : 594 BAYVIEW :2725 UNFOUNDED : 260
## 20:00 : 586 INGLESIDE:2378 JUVENILE BOOKED: 163
## (Other):25098 (Other) :7349 (Other) : 468
## Address X Y
## 800 Block of BRYANT ST : 948 Min. :-122.5 Min. :37.71
## 800 Block of MARKET ST : 288 1st Qu.:-122.4 1st Qu.:37.76
## 900 Block of POTRERO AV : 230 Median :-122.4 Median :37.78
## 1000 Block of POTRERO AV: 199 Mean :-122.4 Mean :37.77
## 2000 Block of MISSION ST: 149 3rd Qu.:-122.4 3rd Qu.:37.79
## 16TH ST / MISSION ST : 116 Max. :-122.4 Max. :37.82
## (Other) :27063
## Location PdId
## (37.775420706711, -122.403404791479) : 940 Min. :1.028e+12
## (37.7571580431915, -122.406604919508): 224 1st Qu.:1.405e+13
## (37.7564864109309, -122.406539115148): 196 Median :1.406e+13
## (37.7650501214668, -122.419671780296): 152 Mean :1.420e+13
## (37.7841893501425, -122.407633520742): 150 3rd Qu.:1.407e+13
## (37.7285280627465, -122.475647460786): 102 Max. :9.904e+13
## (Other) :27229str(SanFran)## 'data.frame': 28993 obs. of 13 variables:
## $ IncidntNum: int 140734311 140736317 146177923 146177531 140734220 140734349 140734349 140734349 140738147 140734258 ...
## $ Category : Factor w/ 34 levels "ARSON","ASSAULT",..: 1 20 16 16 20 7 7 6 21 30 ...
## $ Descript : Factor w/ 368 levels "ABANDONMENT OF CHILD",..: 15 179 143 143 132 247 239 93 107 347 ...
## $ DayOfWeek : Factor w/ 7 levels "Friday","Monday",..: 4 4 4 4 4 4 4 4 4 4 ...
## $ Date : Factor w/ 92 levels "06/01/2014","06/02/2014",..: 92 92 92 92 92 92 92 92 92 92 ...
## $ Time : Factor w/ 1379 levels "00:01","00:02",..: 1370 1365 1351 1351 1344 1334 1334 1334 1321 1321 ...
## $ PdDistrict: Factor w/ 10 levels "BAYVIEW","CENTRAL",..: 1 4 8 7 7 8 8 8 3 2 ...
## $ Resolution: Factor w/ 16 levels "ARREST, BOOKED",..: 12 12 12 12 12 1 1 1 12 2 ...
## $ Address : Factor w/ 8055 levels "0 Block of 10TH ST",..: 6843 4022 1098 6111 5096 1263 1263 1263 1575 5236 ...
## $ X : num -122 -122 -122 -122 -123 ...
## $ Y : num 37.7 37.8 37.8 37.8 37.8 ...
## $ Location : Factor w/ 8732 levels "(37.7080829769301, -122.419241455854)",..: 1970 3730 5834 4802 4777 4993 4993 4993 2543 7598 ...
## $ PdId : num 1.41e+13 1.41e+13 1.46e+13 1.46e+13 1.41e+13 ...head(SanFran)## IncidntNum Category Descript DayOfWeek
## 1 140734311 ARSON ARSON OF A VEHICLE Sunday
## 2 140736317 NON-CRIMINAL LOST PROPERTY Sunday
## 3 146177923 LARCENY/THEFT GRAND THEFT FROM LOCKED AUTO Sunday
## 4 146177531 LARCENY/THEFT GRAND THEFT FROM LOCKED AUTO Sunday
## 5 140734220 NON-CRIMINAL FOUND PROPERTY Sunday
## 6 140734349 DRUG/NARCOTIC POSSESSION OF MARIJUANA Sunday
## Date Time PdDistrict Resolution Address
## 1 08/31/2014 23:50 BAYVIEW NONE LOOMIS ST / INDUSTRIAL ST
## 2 08/31/2014 23:45 MISSION NONE 400 Block of CASTRO ST
## 3 08/31/2014 23:30 SOUTHERN NONE 1000 Block of MISSION ST
## 4 08/31/2014 23:30 RICHMOND NONE FULTON ST / 26TH AV
## 5 08/31/2014 23:23 RICHMOND NONE 800 Block of LA PLAYA ST
## 6 08/31/2014 23:13 SOUTHERN ARREST, BOOKED 11TH ST / MINNA ST
## X Y Location PdId
## 1 -122.4056 37.73832 (37.7383221869053, -122.405646994567) 1.407343e+13
## 2 -122.4350 37.76177 (37.7617677182954, -122.435012093789) 1.407363e+13
## 3 -122.4098 37.78004 (37.7800356268394, -122.409795194505) 1.461779e+13
## 4 -122.4853 37.77252 (37.7725176473142, -122.485262988324) 1.461775e+13
## 5 -122.5099 37.77231 (37.7723131976814, -122.509895418239) 1.407342e+13
## 6 -122.4166 37.77391 (37.773907074489, -122.416578493475) 1.407343e+13Crime Count General
How many different kinds of crimes and how frequent?
crimes = levels(SanFran$Category)
SanFran$Category = sapply(as.character(SanFran$Category),
function(x) ifelse(grepl('THEFT',x),
'THEFT',x))
#SanFran$Category[which(SanFran$Category == 'VEHICLE THEFT')] = 'LARCENY/THEFT'
#SanFran$Category=factor(SanFran$Category,labels=unique(SanFran$Category))I merged together the categories vehicle theft and larceny theft because, looking at some of the descriptions, I noticed that some vehicle theft have been categorized as larceny/theft, so it made no sense to keep the two separate (look at the output of head(SanFran) above).
Now I can check the most occurring crime categories
Cat = group_by(SanFran, Category)
CrimeCount = dplyr::summarise(Cat, count = n())
CrimeByCat=ggplot(data=CrimeCount, aes(x=CrimeCount$Category, y=CrimeCount$count))+
geom_bar(aes(fill=CrimeCount$count),stat="identity") +coord_flip()
CrimeByCat = CrimeByCat+
scale_fill_distiller(name="Nr. of Crimes", palette = "Reds", breaks = pretty_breaks(n = 15),trans='reverse')
#
CrimeByCat = CrimeByCat+ guides(fill=guide_legend(keywidth=0.2,keyheight=0.2,default.unit="inch"))
#
CrimeByCat
The most frequent crimes are related to thefts, followed by non-criminal and other offenses, warrants, vehicle thefts. Missing people and assaults are also non-negligible.
Crime Count by neighborhood
Which is the district with the highest amount of crimes?
Let’s see how crimes are distributed on the surface of San Francisco
names(SanFran)[which(names(SanFran)=='X')] = 'Longitude'
names(SanFran)[which(names(SanFran)=='Y')] = 'Latitude'
xrange = max(SanFran$Longitude)-min(SanFran$Longitude)
yrange = max(SanFran$Latitude)-min(SanFran$Latitude)
xlim = c(min(SanFran$Longitude),max(SanFran$Longitude))
ylim = c(min(SanFran$Latitude),max(SanFran$Latitude))
crimes_per_d = ddply(SanFran,.(PdDistrict),nrow)
names(crimes_per_d)[2] = 'TotalNumCrimes'
SanFranmap = ddply(SanFran,.(PdDistrict),summarize,
meanLat = round(mean(Latitude),2),
meanLong = round(mean(Longitude),2))
SanFranmap['TotNumCrimes'] = crimes_per_d['TotalNumCrimes']
set.seed(8000)
#http://www.r-bloggers.com/mapping-with-ggplot-create-a-nice-choropleth-map-in-r/
#https://data.sfgov.org/Public-Safety/Historic-Police-Districts/embj-38bg
PdDistricts.shp <- readShapeSpatial("./Historic\ Police\ Districts//geo_export_061708fa-db2a-48c9-8e51-f2548a086820.shp")
v = PdDistricts.shp$objectid
names(v) = PdDistricts.shp$district
SanFranmap$id = sapply(SanFranmap$PdDistrict, function(x) v[as.character(x)])
PdDistricts.f = fortify(PdDistricts.shp, region = "objectid")
merge.shp.coef<-merge(PdDistricts.f, SanFranmap, by="id", all.x=TRUE)
final.plot<-merge.shp.coef[order(merge.shp.coef$order), ]
SanFranmap$Latrev = SanFranmap$meanLat
SanFranmap$Longrev = SanFranmap$meanLong
SanFranmap$Latrev[8] = 37.777
SanFranmap$Longrev[8] = -122.395
SanFranmap$Latrev[10] = 37.785
g = ggplot() +
geom_polygon(data = final.plot,
aes(x = long, y = lat, group = group, fill = TotNumCrimes),
color = "black", size = 0.25) +
coord_map() +
scale_fill_distiller(name="Nr. of Crimes", palette = "Reds", breaks = pretty_breaks(n = 5),trans='reverse')+
theme_nothing(legend = TRUE)+
labs(title="Total nr. of crimes per Police District: San Francisco")+
geom_text(data=SanFranmap, aes(Longrev, Latrev, label = PdDistrict), size=5, fontface="bold")
g
The district with the highest number of crimes is Southern and in general crimes peak on the Eastern coastal districts, decreasing as one goes towards the West
Do kinds of crime vary by district?
Top 5 most occurring crimes by district
CrimeCount=CrimeCount[ order(-CrimeCount[,2]), ]
Top5 = as.character(CrimeCount$Category[1:5])
SanFran$Top5 = sapply(SanFran$Category, function(x) ifelse (as.character(x) %in% Top5,as.character(x),'Others'))
bycat = group_by(SanFran,Top5,PdDistrict)
sumc = dplyr::summarise(bycat,count = n())
ggplot(sumc, aes(x = PdDistrict, y = count, fill = Top5)) +
geom_bar(stat = 'identity', position = 'fill', colour = 'black') +
#facet_wrap(~PdDistrict) +
coord_flip() +
labs(y = 'Percentage of crimes',
x = 'Police District',
title = 'Top 5 most occurring crimes by District') 
In Southern, Central and Northern district the relative majority of thefts is more pronounced than in the other districts where the crime-type distribution seems more diversified with Mission, Tenderloin and Bayview being the most diversified with the lowest percentage of thefts.
Thefts by district and hour of day
SanFran$Date = as.POSIXct(SanFran$Date, format="%m/%d/%Y")
SanFran$Month = month(SanFran$Date)
SanFran$Day = day(SanFran$Date)
SanFran$Time = as.POSIXct(SanFran$Time, format="%H:%M")
SanFran$Hour = hour(SanFran$Time)
subs = SanFran[which(SanFran$Category=='THEFT'),]
by_cat = group_by(subs,PdDistrict,Hour)
sumc = dplyr::summarise(by_cat,count = n())
ggplot(sumc, aes(x = PdDistrict, y = Hour)) +
geom_tile(aes(fill = count)) +
scale_fill_distiller(name="Nr. of Thefts", palette = "Reds", breaks = pretty_breaks(n = 5),trans='reverse')+
theme(axis.title.y = element_blank()) + theme_light(base_size = 10) +
theme(plot.title = element_text(size = 16)) +
ggtitle("Theft count by District and hour") +
theme(axis.text.x = element_text(angle = 45,size = 8, vjust = 0.5)) 
We see that, as already mentioned, the Southern, Central and Northern district exhibit a much more pronounced theft pattern. Thefts seem more likely to occurr in the afternoon-evening alotough, in the Southern district, one should watch out all day long except in the earliest hours of the day