-
Notifications
You must be signed in to change notification settings - Fork 0
/
classification.Rmd
225 lines (176 loc) · 5.72 KB
/
classification.Rmd
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
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
---
title: "Obesity"
author: "Kristian Ekachandra"
date: "2021"
output: openintro::lab_report
---
#Load all packages
```{r load-packages, message=FALSE, warning=FALSE}
library(Amelia)
library(ggplot2)
library(GGally)
library(tidyverse)
library(knitr)
library(rpart)
library(rpart.plot)
library(party)
library(caret)
```
#Data Validation
```{r, message=FALSE, warning=FALSE}
#Read data from csv
dat <- read.csv("dataset.csv")
str(dat)
summary(dat)
#Data Validation
missmap(dat)
#data type into factor
dat$Gender <-as.factor(dat$Gender)
dat$family_history_with_overweight <-as.factor(dat$family_history_with_overweight)
dat$FAVC<-as.factor(dat$FAVC)
dat$CAEC<-as.factor(dat$CAEC)
dat$SCC <-as.factor(dat$SCC)
dat$SMOKE<-as.factor(dat$SMOKE)
dat$CALC<-as.factor(dat$CALC)
dat$MTRANS<-as.factor(dat$MTRANS)
dat$NObeyesdad<-as.factor(dat$NObeyesdad)
#Sort to the right order
dat$CAEC <- factor(dat$CAEC, levels = c("Always", "Frequently", "Sometimes", "no"))
dat$CALC <- factor(dat$CALC, levels = c("Always", "Frequently", "Sometimes", "no"))
dat$MTRANS <- factor(dat$MTRANS, levels = c("Walking", "Bike", "Motorbike", "Automobile", "Public_Transportation"))
dat$NObeyesdad <- factor(dat$NObeyesdad, levels = c("Insufficient_Weight", "Normal_Weight", "Overweight_Level_I", "Overweight_Level_II", "Obesity_Type_I", "Obesity_Type_II", "Obesity_Type_III"))
```
#Data Visualitation
```{r, message=FALSE, warning=FALSE}
#Data Categorical
##Gender
ggplot2::ggplot(data = dat) +
aes(x = Gender, fill = Gender) +
geom_bar() +
labs(title = "Distribution by Gender")
##Family History With Overweight
ggplot2::ggplot(data = dat) +
aes(x = family_history_with_overweight, fill = family_history_with_overweight) +
geom_bar() +
labs(title = "Distribution by Obesity Class Family History With Overweight")
##Frequent consumption of high caloric food
ggplot2::ggplot(data = dat) +
aes(x = FAVC, fill = FAVC) +
geom_bar() +
labs(title = "Distribution by Frequent consumption of high caloric food")
##Consumption of food between meals
ggplot2::ggplot(data = dat) +
aes(x = CAEC, fill = CAEC) +
geom_bar() +
labs(title = "Distribution by Consumption of food between meals")
##SMOKE
ggplot2::ggplot(data = dat) +
aes(x = SMOKE, fill = SMOKE) +
geom_bar() +
labs(title = "Distribution by SMOKE")
##Calories consumption monitoring
ggplot2::ggplot(data = dat) +
aes(x = SCC, fill = SCC) +
geom_bar() +
labs(title = "Distribution by Calories consumption monitoring")
##Consumption of alcohol
ggplot2::ggplot(data = dat) +
aes(x = CALC, fill = CALC) +
geom_bar() +
labs(title = "Distribution by Consumption of alcohol")
##Transportation used
ggplot2::ggplot(data = dat) +
aes(x = MTRANS, fill = MTRANS) +
geom_bar() +
labs(title = "Distribution by Transportation used") +
theme(axis.text.x = element_text(size = 10, angle = 20))
##Obesity
ggplot2::ggplot(data = dat) +
aes(x = NObeyesdad, fill = NObeyesdad) +
geom_bar() +
labs(title = "Distribution by Obesity Class") +
theme(axis.text.x = element_text(size = 10, angle = 20))
#Data Numeric
##Age
ggplot2::ggplot(data = dat,aes(Age)) +
geom_density(fill="light blue", color="light blue", alpha=0.8)+
ggtitle("Distribution of Age") +
theme_classic()
#Corelation
library(GGally)
ggcorr(dat, method = c("everything"))+
labs(title = "Predictor Variables")
```
#Split Data
```{r, message=FALSE, warning=FALSE}
NIM <- 43961
set.seed(NIM)
samp <- sample(nrow(dat), 0.8 * nrow(dat), replace = FALSE)
data_train <- dat[samp, ] #80% data_train
data_test <- dat[-samp, ] #20% data_test
```
#Algorithm :
1. Decision Tree
```{r fig, message=FALSE, warning=FALSE, fig.height=20, fig.width=15}
#rpart
#fit model
obesity_rpart <- rpart(NObeyesdad ~.,
data = data_train,
method = "class")
#plot
rpart.plot(obesity_rpart,
main= "Obesity Data with Decision Tree (rpart)",
space = 0,
split.cex = 2,
nn.border.col = 4,
box.palette="RdBu", nn = TRUE)
print(obesity_rpart)
#predict
predict_rpart <- predict(obesity_rpart, data_test, type = "class")
#confusion matrix rpart
caret::confusionMatrix(predict_rpart, data_test$NObeyesdad)
# Accuracy : 0.8345
#party
#fit model
obesity_party <- ctree(NObeyesdad ~ ., data = data_train)
#plot
plot(obesity_party,
type = "simple",
main = "Obesity Data with Decision Tree (party)")
print(obesity_party)
#predict
predict_party <- predict(obesity_party, data_test, type = "response")
(table_party = table(predict_party,data_test$NObeyesdad))
#confusion matrix party
caret::confusionMatrix(table_party)
# Accuracy : 0.9196
#the comparison of accuracy using party (91.96%) is higher than using rpart (83.45%).
```
2. Naive Bayes
```{r, message=FALSE, warning=FALSE}
#making model
nb_mod <- klaR::NaiveBayes(NObeyesdad ~ Gender + Age + Height + Weight + family_history_with_overweight + FAVC + CAEC + SCC + SMOKE + CALC + MTRANS + CH2O, data = data_train)
#predict
pred <- predict(nb_mod, data_test) #test model to data testing
#confusion matrix
(tab <- table(pred$class, data_test$NObeyesdad))
caret::confusionMatrix(tab)
#Accuracy : 70.69%
#plot
data_test$pred <- pred$class
Xket <- "Truth"
Yket <- "Predicted"
Judul <- "Obesity - Naive Bayess"
ggplot(data = data_test) +
aes(NObeyesdad, pred, color = NObeyesdad) +
geom_jitter(width = 0.2, height = 0.1, size = 3) +
labs(title = Judul,
subtitle = "Predicted vs. Observed from Obesity dataset",
x = Xket, y = Yket) +
theme(axis.text.x = element_text(size = 10, angle = 20))
```
#Conclusion
Using the decision tree algorithm in classifying obesity data is better than using the Naive Bayes algorithm.
By comparison accuracy:
1. Decision Tree (party) : 91.96%
2. Naive Bayes : 70.69%