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---
title: "stat445project"
output: html_document
---
```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```
Install packages needed.
```{r}
install.packages("caret")
install.packages("ggplot2")
library(caret)
library(ggplot2)
```
## Importing the datasets
```{r}
setwd("/Users/justinlew/Documents/SFU/STAT445/project/pokemon-challenge")
getwd()
combats <- read.csv("combats.csv")
pokemon <- read.csv("pokemon.csv")
tests <- read.csv("tests.csv")
combats2 <- read.csv("combats.csv")
tests
head(combats)
head(pokemon)
head(tests)
combats$HP <- NA
combats$Attack <- NA
combats$Defense <- NA
combats$Sp..Atk <- NA
combats$Sp..Def <- NA
combats$Speed <- NA
combats$Total <- NA
combats$Result <- NA
for (row in 1:nrow(combats)) {
#if the first pokemon won
if (combats[row, 3] == combats[row, 1]) {
combats[row, ]$HP <- pokemon[pokemon$X.==combats[row, 3],]$HP - pokemon[pokemon$X.==combats[row, 2],]$HP
combats[row, ]$Attack <- pokemon[pokemon$X.==combats[row, 3],]$Attack - pokemon[pokemon$X.==combats[row, 2],]$Attack
combats[row, ]$Defense <- pokemon[pokemon$X.==combats[row, 3],]$Defense - pokemon[pokemon$X.==combats[row, 2],]$Defense
combats[row, ]$Sp..Atk <- pokemon[pokemon$X.==combats[row, 3],]$Sp..Atk - pokemon[pokemon$X.==combats[row, 2],]$Sp..Atk
combats[row, ]$Sp..Def <- pokemon[pokemon$X.==combats[row, 3],]$Sp..Def - pokemon[pokemon$X.==combats[row, 2],]$Sp..Def
combats[row, ]$Speed <- pokemon[pokemon$X.==combats[row, 3],]$Speed - pokemon[pokemon$X.==combats[row, 2],]$Speed
combats[row, ]$Result <- 0
} else {
#if the second pokemon won
combats[row, ]$HP <- pokemon[pokemon$X.==combats[row, 3],]$HP - pokemon[pokemon$X.==combats[row, 1],]$HP
combats[row, ]$Attack <- pokemon[pokemon$X.==combats[row, 3],]$Attack - pokemon[pokemon$X.==combats[row, 1],]$Attack
combats[row, ]$Defense <- pokemon[pokemon$X.==combats[row, 3],]$Defense - pokemon[pokemon$X.==combats[row, 1],]$Defense
combats[row, ]$Sp..Atk <- pokemon[pokemon$X.==combats[row, 3],]$Sp..Atk - pokemon[pokemon$X.==combats[row, 1],]$Sp..Atk
combats[row, ]$Sp..Def <- pokemon[pokemon$X.==combats[row, 3],]$Sp..Def - pokemon[pokemon$X.==combats[row, 1],]$Sp..Def
combats[row, ]$Speed <- pokemon[pokemon$X.==combats[row, 3],]$Speed - pokemon[pokemon$X.==combats[row, 1],]$Speed
combats[row, ]$Result <- 1
}
}
combats$Result <- factor(combats$Result)
logisticRegression <- glm(Result ~ HP + Attack + Defense + Sp..Atk + Sp..Def + Speed, data = combats, family = "binomial")
summary(logisticRegression)
predict(logisticRegression, newdata = test)
```
## Manipulating the test dataset
```{r}
calcDiff <- function(arr) {
counter <- 0
arr$HP <- NA
arr$Attack <- NA
arr$Defense <- NA
arr$Sp..Atk <- NA
arr$Sp..Def <- NA
arr$Speed <- NA
arr$Total <- NA
arr$Result <- NA
for (row in 1:nrow(arr)) {
arr[row, ]$HP <- pokemon[pokemon$X.==arr[row, 1],]$HP - pokemon[pokemon$X.==arr[row, 2],]$HP
arr[row, ]$Attack <- pokemon[pokemon$X.==arr[row, 1],]$Attack - pokemon[pokemon$X.==arr[row, 2],]$Attack
arr[row, ]$Defense <- pokemon[pokemon$X.==arr[row, 1],]$Defense - pokemon[pokemon$X.==arr[row, 2],]$Defense
arr[row, ]$Sp..Atk <- pokemon[pokemon$X.==arr[row, 1],]$Sp..Atk - pokemon[pokemon$X.==arr[row, 2],]$Sp..Atk
arr[row, ]$Sp..Def <- pokemon[pokemon$X.==arr[row, 1],]$Sp..Def - pokemon[pokemon$X.==arr[row, 2],]$Sp..Def
arr[row, ]$Speed <- pokemon[pokemon$X.==arr[row, 1],]$Speed - pokemon[pokemon$X.==arr[row, 2],]$Speed
if (arr[row, 1] == arr[row, 3]) {
#result is 1 if the first Pokemon who attacks first wins
arr[row, ]$Result <- 1
} else {
arr[row, ]$Result <- 0
}
if (row %% 1000 == 0) {
counter <- counter + 1
print(counter)
}
}
return(arr)
}
combats <- calcDiff(combats)
```
Split into training and test set.
```{r}
sampleSize <- floor(0.75 * nrow(combats))
set.seed(123)
train_indices <- sample(seq_len(nrow(combats)), size = sampleSize)
train <- combats[train_indices,]
test <- combats[-train_indices,]
```
```{r}
```
Perform logistic regression on training set.
```{r}
install.packages("e1071")
library(e1071)
train$Result <- factor(train$Result)
test$Result <- factor(test$Result)
logisticRegression <- glm(Result ~ HP + Attack + Defense + Sp..Atk + Sp..Def + Speed, data = train, family = "binomial")
summary(logisticRegression)
logisticPrediction <- round(predict(logisticRegression, newdata = test, type="response"))
confusionMatrix(logisticPrediction, test$Result)
test$Sample <- logisticPrediction
ggplot(test, aes(x=Sp..Def, y=Sample)) + geom_point() + stat_smooth(method="glm", method.args=list(family="binomial"), se=FALSE)
```
```{r}
install.packages("class")
library(class)
```
```{r}
training <- train
testing <- test
training$Total <- NULL
testing$Total <- NULL
training$Result <- NULL
training$First_pokemon <- NULL
training$Second_pokemon <- NULL
training$Winner <- NULL
testing$Result <- NULL
testing$First_pokemon <- NULL
testing$Second_pokemon <- NULL
testing$Winner <- NULL
knn <- knn(train=training, test=testing, cl = train$Result, k=2, prob=TRUE)
sum(knn == test$Result)/12500
table(knn, test$Result)
```