CoCalc -- kernel

GitHub Repository: debakarr/machinelearning
Path: blob/master/Part 9 - Dimension Reduction/Kernel PCA/kernel_pca.R
¹³³⁵ views
1
# Kernel PCA
2

3
# Importing the dataset
4
dataset = read.csv('Social_Network_Ads.csv')
5
dataset = dataset[, 3:5]
6

7
# Splitting the dataset into the Training set and Test set
8
# install.packages('caTools')
9
library(caTools)
10
set.seed(123)
11
split = sample.split(dataset$Purchased, SplitRatio = 0.75)
12
training_set = subset(dataset, split == TRUE)
13
test_set = subset(dataset, split == FALSE)
14

15
# Feature Scaling
16
training_set[, 1:2] = scale(training_set[, 1:2])
17
test_set[, 1:2] = scale(test_set[, 1:2])
18

19
# Applying Kernel PCA
20
# install.packages('kernlab')
21
library(kernlab)
22
kpca = kpca(~., data = training_set[-3], kernel = 'rbfdot', features = 2)
23
training_set_pca = as.data.frame(predict(kpca, training_set))
24
training_set_pca$Purchased = training_set$Purchased
25
test_set_pca = as.data.frame(predict(kpca, test_set))
26
test_set_pca$Purchased = test_set$Purchased
27

28
# Fitting Logistic Regression to the Training set
29
classifier = glm(formula = Purchased ~ .,
30
                 family = binomial,
31
                 data = training_set_pca)
32

33
# Predicting the Test set results
34
prob_pred = predict(classifier, type = 'response', newdata = test_set_pca[-3])
35
y_pred = ifelse(prob_pred > 0.5, 1, 0)
36

37
# Making the Confusion Matrix
38
cm = table(test_set_pca[, 3], y_pred)
39

40
# Visualising the Training set results
41
install.packages('ElemStatLearn')
42
library(ElemStatLearn)
43
set = training_set_pca
44
X1 = seq(min(set[, 1]) - 1, max(set[, 1]) + 1, by = 0.01)
45
X2 = seq(min(set[, 2]) - 1, max(set[, 2]) + 1, by = 0.01)
46
grid_set = expand.grid(X1, X2)
47
colnames(grid_set) = c('V1', 'V2')
48
prob_set = predict(classifier, type = 'response', newdata = grid_set)
49
y_grid = ifelse(prob_set > 0.5, 1, 0)
50
plot(set[, -3],
51
     main = 'Logistic Regression (Training set)',
52
     xlab = 'PC1', ylab = 'PC2',
53
     xlim = range(X1), ylim = range(X2))
54
contour(X1, X2, matrix(as.numeric(y_grid), length(X1), length(X2)), add = TRUE)
55
points(grid_set, pch = '.', col = ifelse(y_grid == 1, 'springgreen3', 'tomato'))
56
points(set, pch = 21, bg = ifelse(set[, 3] == 1, 'green4', 'red3'))
57

58
# Visualising the Test set results
59
# install.packages('ElemStatLearn')
60
library(ElemStatLearn)
61
set = test_set_pca
62
X1 = seq(min(set[, 1]) - 1, max(set[, 1]) + 1, by = 0.01)
63
X2 = seq(min(set[, 2]) - 1, max(set[, 2]) + 1, by = 0.01)
64
grid_set = expand.grid(X1, X2)
65
colnames(grid_set) = c('V1', 'V2')
66
prob_set = predict(classifier, type = 'response', newdata = grid_set)
67
y_grid = ifelse(prob_set > 0.5, 1, 0)
68
plot(set[, -3],
69
     main = 'Logistic Regression (Test set)',
70
     xlab = 'Age', ylab = 'Estimated Salary',
71
     xlim = range(X1), ylim = range(X2))
72
contour(X1, X2, matrix(as.numeric(y_grid), length(X1), length(X2)), add = TRUE)
73
points(grid_set, pch = '.', col = ifelse(y_grid == 1, 'springgreen3', 'tomato'))
74
points(set, pch = 21, bg = ifelse(set[, 3] == 1, 'green4', 'red3'))
75
Product

Resources

Company
