-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathknn.R
230 lines (208 loc) · 5.2 KB
/
knn.R
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
226
227
228
229
230
# KNN 41.53654% after scaling its 34.25%
library(class)
train <- read.csv(file="training.csv")
test <- read.csv(file="test.csv")
# Correct the data types
train$relevance <- as.factor(train$relevance)
# train$sig3 = log(train$sig3+1)
# train$sig4 = log(train$sig4+1)
# train$sig5 = log(train$sig5+1)
# Because the KNN classifier predicts the class of a given test observation by
# identifying the observations that are nearest to it, the scale of the variables
# matters. Any variables that are on a large scale will have a much larger
# effect on the distance between the observations, and hence on the KNN
# classifier, than variables that are on a small scale.
set.seed(2325)
train_indices = sample(1:dim(train)[1], dim(train)[1]*0.8)
aTrain <- train[train_indices, ]
vTrain <- train[-train_indices, ]
# A matrix containing the predictors associated with the data for train and test
train.X = scale(cbind(aTrain[,3:12]))
test.X = scale(cbind(vTrain[,3:12]))
# A vector containing the class labels for the training observations
train.relevance = aTrain$relevance
kSet = c(1,2,10,50,100,150,200,250,300,350,400,450)
train.err <- rep(NA, length(kSet))
#Tuning for k using LOOCV
# for(i in 1:length(kSet)){
# loocv.pred = knn.cv(train.X,train.relevance,k=kSet[i])
# table(loocv.pred,aTrain$relevance)
# cat("\n\n Error rate for KNN ", mean(loocv.pred!=aTrain$relevance)*100)
# cat("\nk = ",kSet[i])
# }
# A value for K, the number of nearest neighbors to be used by the classifier (found by cv)
k = 200
set.seed(2325)
knn.pred = knn(train.X, test.X, train.relevance, k)
conf = table(knn.pred,vTrain$relevance)
plot(conf)
cat("\n\n Error rate for KNN ", mean(knn.pred!=vTrain$relevance)*100)
# LOOCV, k = 300 is the optimal value
loocv.err = c(42.31214, 42.13411, 37.40396, 34.54463, 34.23699, 34.09488, 34.1308, 34.11362, 34.08551, 34.11206, 34.16047, 34.12924)
kSet = c(1, 2, 10, 50, 100, 150, 200, 250, 300, 350, 400, 450)
plot(kSet,loocv.err,type ="b")
# Error rate for KNN 42.31214
# k = 1
#
# Error rate for KNN 42.13411
# k = 2
#
# Error rate for KNN 37.40396
# k = 10
#
# Error rate for KNN 34.54463
# k = 50
#
# Error rate for KNN 34.23699
# k = 100
#
# Error rate for KNN 34.09488
# k = 150
#
# Error rate for KNN 34.1308
# k = 200
#
# Error rate for KNN 34.11362
# k = 250
#
# Error rate for KNN 34.08551
# k = 300
#
# Error rate for KNN 34.11206
# k = 350
#
# Error rate for KNN 34.16047
# k = 400
#
# Error rate for KNN 34.12924
# k = 450
# RESULTS - before scaling
# k = 1 Error rate for KNN 46.05871
# k = 2 Error rate for KNN 46.35853
# k = 3 Error rate for KNN 45.03435
# k = 4 Error rate for KNN 45.58401
# k = 5 Error rate for KNN 43.8476
# k = 6 Error rate for KNN 44.36602
# k = 7 Error rate for KNN 43.59151
# k = 8 Error rate for KNN 44.00999
# k = 9 Error rate for KNN 43.49781
# k = 10 Error rate for KNN 43.46034
# k = 11 Error rate for KNN 43.34791
# k = 12 Error rate for KNN 43.0356
# k = 13 Error rate for KNN 42.79825
# k = 14 Error rate for KNN 43.01062
# k = 15 Error rate for KNN 42.56715
# k = 16 Error rate for KNN 42.75453
# k = 17 Error rate for KNN 42.38601
# k = 18 Error rate for KNN 42.69831
# k = 19 Error rate for KNN 42.40475
# k = 20 Error rate for KNN 42.49844
# After scaling
# Error rate for KNN 42.29856 - 1
# Error rate for KNN 43.37914 - 2
# Error rate for KNN 39.05059 - 5
# Error rate for KNN 37.83885 - 10
# Error rate for KNN 35.64022 - 20
# Error rate for KNN 34.57214 - 50
# Error rate for KNN 34.20987 - 100
# Error rate for KNN 34.26608 - 150
# Error rate for KNN 34.23485 - 200
# (200,250,300,350,400,450)
# Error rate for KNN 34.27858
#
# Error rate for KNN 34.14116
#
# Error rate for KNN 34.25359
#
# Error rate for KNN 34.35978
#
# Error rate for KNN 34.36602
#
# Error rate for KNN 34.22861
# Cross validation - c(1,2,5,10,15,40,70,100,150,200,300,400,500,600,700)
# Error rate for KNN 42.37304
#
# Error rate for KNN 42.38709
#
# Error rate for KNN 38.63296
#
# Error rate for KNN 37.30558
#
# Error rate for KNN 35.97664
#
# Error rate for KNN 34.68674
#
# Error rate for KNN 34.45562
#
# Error rate for KNN 34.31507
#
# Error rate for KNN 34.14642
#
# Error rate for KNN 34.08551
#
# Error rate for KNN 34.0293
#
# Error rate for KNN 34.0699
# CV of KNN in 300 to 400
# Error rate for KNN 34.00587
# k = 300
#
# Error rate for KNN 34.03554
# k = 305
#
# Error rate for KNN 34.0293
# k = 310
#
# Error rate for KNN 33.99963
# k = 315
#
# Error rate for KNN 34.04804
# k = 320
#
# Error rate for KNN 33.99963
# k = 325
#
# Error rate for KNN 34.03086
# k = 330
#
# Error rate for KNN 34.02773
# k = 335
#
# Error rate for KNN 34.05428
# k = 340
#
# Error rate for KNN 34.01836
# k = 345
#
# Error rate for KNN 34.01993
# k = 350
#
# Error rate for KNN 34.05741
# k = 355
#
# Error rate for KNN 34.03867
# k = 360
#
# Error rate for KNN 34.08864
# k = 365
#
# Error rate for KNN 34.09488
# k = 370
#
# Error rate for KNN 34.09801
# k = 375
#
# Error rate for KNN 34.08864
# k = 380
#
# Error rate for KNN 34.11831
# k = 385
#
# Error rate for KNN 34.08551
# k = 390
#
# Error rate for KNN 34.07302
# k = 395
#
# Error rate for KNN 34.08239
# k = 400