Visualizations-and-Classifier-of-Crimes-in-Liverpool-Area/R Script_Visualizations and Classifier of Crimes in the Liverpool Area.R at main · leemnto/Visualizations-and-Classifier-of-Crimes-in-Liverpool-Area · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
#Install Packages
install.packages("tidyverse")
install.packages("tmap")
install.packages("caret")
install.packages("rpart")
install.packages("cartogram")
install.packages("missForest")
install.packages("randomForest")
install.packages("dplyr")
install.packages("base")
install.packages("GGally")

library(tidyverse)
library(tmap)
library(caret)
library(rpart)
library(cartogram)
library(missForest)
library(randomForest)
library(dplyr)
library(base)
library(GGally)


#Read Liverpool mapfile
liverpoolshape<-readOGR(dsn="./BoundaryData", layer="england_lsoa_2011")
qtm(liverpoolshape)


#Read Population Data
deprivation2015<-read.csv("2015ranks.csv")

#Transform Population Data and Rename Columns
deprivation2015code<-deprivation2015 %>% select(LSOA.name..2011.,
                                                LSOA.code..2011.,
                                                Total.population..mid.2012..excluding.prisoners.)
names(deprivation2015code)[names(deprivation2015code)=="LSOA.name..2011."]<-"LSOA_name"
names(deprivation2015code)[names(deprivation2015code)=="LSOA.code..2011."]<-"LSOA_code"
names(deprivation2015code)[names(deprivation2015code)=="Total.population..mid.2012..excluding.prisoners."]<-"Total_population"

#Match and Combine Population Data and Map File Data
liverpoolshape@data<-left_join(liverpoolshape@data, deprivation2015code,
                               by=c('code'='LSOA_code'))

#Create Population Distribution Map
qtm(liverpoolshape, fill="Total_population", fill.palette="Blues", borders="White")

#Read Crime Data
crimes<-read.csv("1911street.csv", header=TRUE)

#Create Bar Chart of Crime Types
typecrimes<-crimes %>%
  select(LSOA.code, Crime.type) %>%
  group_by(Crime.type) %>%
  summarise(typecrimes=n())
ggplot(typecrimes, aes( x = reorder(Crime.type, typecrimes),
                        y = typecrimes)) + geom_col() +
  labs(x="Crime Type", y="Number of Crimes",title = "Crimes in 2019-11") +
  coord_flip() +
  theme_light()

#Transform Crime Data
noofcrimes<-crimes %>%
  select(LSOA.code, LSOA.name, Crime.type) %>%
  group_by(LSOA.code) %>%
  summarise(nocrimes=n())

#Match and Combine Crime Data and Map File Data
liverpoolshape@data<-left_join(liverpoolshape@data, noofcrimes,
                               by=c('code'='LSOA.code'))
#Transform Missing Data
liverpoolshape[is.na(liverpoolshape@data$nocrimes)]<-0

#Create Crime Distribution Map
tm_shape(liverpoolshape) +
  tm_fill("nocrimes", style="kmeans", border.col = "black", title = "Number of Crimes") +
  tm_borders(alpha=0.2)

#Create Crime Distribution Cartogram
livercar<-cartogram(liverpoolshape, weight="nocrimes", itermax=10, prepare="adjust")
tm_shape(livercar) +
  tm_fill("nocrimes", style="jenks", title = "Number of Crimes") +
  tm_borders() + tm_layout(frame=F)

#Test for Normality - Number of Crimes
noofcrimes<-crimes %>%
  select(LSOA.code, LSOA.name, Crime.type) %>%
  group_by(LSOA.code) %>%
  summarise(nocrimes=n())
shapiro.test(noofcrimes$nocrimes)

#Test for Normality - Population
shapiro.test(liverpoolshape@data$Total_population)

#Correlation Scatter Plot
ggplot(liverpoolshape@data, aes(Total_population, nocrimes)) +
  geom_point()+
labs(x="Total Population",y="Number of Crimes")
#Correlation Test
pop = liverpoolshape@data$Total_population
crime = liverpoolshape@data$nocrimes
cor(pop, crime, method="kendall")

######### Machine Learning ############
#Read Liverpool mapfile
liverpoolshape<-readOGR(dsn="./BoundaryData", layer="england_lsoa_2011")
#Read Population Data
de2015<-read.csv("deprivationdata.csv")
#Match and Combine Data
liverpoolshape@data<-left_join(liverpoolshape@data, de2015,
                               by=c('code'='LSOA.code..2011.'))
liverpoolshape@data<-left_join(liverpoolshape@data, noofcrimes,
                               by=c('code'='LSOA.code'))

#Rename Columns
names(liverpoolshape@data)[names(liverpoolshape@data)=="Crime.Score"]<-"Crime_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Education..Skills.and.Training.Score"]<-"Education_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Index.of.Multiple.Deprivation..IMD..Score"]<-"IMD_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Living.Environment.Score"]<-"Live_Environ_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Adult.Skills.Sub.domain.Score"]<-"Adult_Skill_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Barriers.to.Housing.and.Services.Score"]<-"Barriers_HS_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Dependent.Children.aged.0.15..mid.2012..excluding.prisoners."]<-"De_Children"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Employment.Score..rate."]<-"Employment_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Geographical.Barriers.Sub.domain.Score"]<-"Geo_Bar_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Health.Deprivation.and.Disability.Score"]<-"Health_Dis_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Income.Deprivation.Affecting.Older.People..IDAOPI..Score..rate."]<-"Income_Aff_Old"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Income.Score..rate."]<-"Income_Score_Rate"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Indoors.Sub.domain.Score"]<-"Indoors_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Older.population.aged.60.and.over..mid.2012..excluding.prisoners."]<-"Older_Pop"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Outdoors.Sub.domain.Score"]<-"Outdoors_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Population.aged.16.59..mid.2012..excluding.prisoners."]<-"Middle_Pop"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Working.age.population.18.59.64..for.use.with.Employment.Deprivation.Domain..excluding.prisoners."]<-"Working_Pop"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Children.and.Young.People.Sub.domain.Score"]<-"Young_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Income.Deprivation.Affecting.Children.Index..IDACI..Score..rate."]<-"Income_Aff_Child"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Wider.Barriers.Sub.domain.Score"]<-"Wider_Bar_Score"
names(liverpoolshape@data)[names(liverpoolshape@data)=="Total.population..mid.2012..excluding.prisoners."]<-"Total_population"

#Independent Values Correlation Plot
livercor = subset(liverpoolshape@data, select = c(7,10,13,16,19,22,25,28,31,34,37,40,43,46,49,52,55,56,57,58,59) )
ggpairs(livercor)

#Deal with Missing Value
randomcrime = subset(liverpoolshape@data, select = c(7,10,13,16,19,22,25,28,31,34,37,40,43,46,49,52,55,56,57,58,59,60))
anyNA(randomcrime)
#Using Random Forest to Predict Missing Values
filter.randomcrime<-missForest(randomcrime)
str(filter.randomcrime)
pima.filter.crime<-filter.randomcrime$ximp
print(table(randomcrime$nocrimes))
print(table(pima.filter.crime$nocrimes))

#Scale and Normalize Data
trans<-pima.filter.crime %>% select(nocrimes)
pima.scale.data<-scale(pima.filter.crime[,-22])
pima.scale.data<-cbind(pima.scale.data, trans)

#Break the Dependent Value (Number of Crimes) into 3 types - high, medium, low
summary(pima.scale.data$nocrimes)
brks=c(0,8,25,466)
pima.scale.data$nocrimes=cut(pima.scale.data$nocrimes, breaks=brks)
str(pima.scale.data$nocrimes)
is.na(pima.scale.data$nocrimes)

#Split the data into Training Dataset and Test Dataset
set.seed(203)
splitdata<-pima.scale.data$nocrimes %>%
  createDataPartition(p=0.8, list=FALSE)
train.data<-pima.scale.data[splitdata,]
test.data<-pima.scale.data[-splitdata,]
#Dataset Overview
summary(train.data)
summary(test.data)
print(table(train.data$nocrimes))
print(table(test.data$nocrimes))

####### Decision Tree #######
#Create Decision Tree
rpart.model<-rpart(nocrimes~., data=train.data, method = "class")
rpart.predictions<-rpart.model %>% predict(test.data, type = "class")
mean(rpart.predictions == test.data$nocrimes)
#Create Decision Tree Plot
plot(rpart.model, main="Classification Tree for the Number of Crimes (Training Data)")
text(rpart.model, digits=45, cex=0.9)

######## Random Forest Classification #######
rfmodel<-randomForest(nocrimes ~.,ntree=2000, data = train.data)
print(rfmodel)

#Variables Importance Sequence
varImpPlot(rfmodel, sort=T, n.var = 21, main = "Variable Importance Plot")

#Train the dataset
rf.predictions<-predict(rfmodel, train.data)
print(caret::confusionMatrix(data = rf.predictions,
                             reference = train.data$nocrimes))
#Predict the Number of Crimes
rf.predictions <- predict(rfmodel, test.data)
print(caret::confusionMatrix(data = rf.predictions, reference = test.data$nocrimes))