Statistics-R/Machine Learning.R at main · MarvinEdorh/Statistics-R · 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
rm(list = ls());
setwd("C:/Users/marvin.edorh_artefac/Desktop/Formation R")
library(dplyr)
install.packages("rpart.plot")
################################################ REGRESSION LINEAIRE ################################################

DATA <- read.csv("quanti.csv", sep=",",header = TRUE)

id <- DATA[,1]

DATA <-  DATA[,2:11]

DATA_ML <- data.frame( DATA$Users, DATA$Page_views, DATA$Campaign, DATA$Transactions)

rownames(DATA_ML) <- id

#install.packages("tidyverse")
library(tidyverse)

glimpse(DATA_ML)

plot(DATA_ML)

model <- lm( DATA.Transactions ~ DATA.Users , data = DATA_ML)
#Ce mod?le est-il bon ?
summary(model)$r.squared
## [1] 0.9419583

new_data <- data.frame(DATA.Users = c(100, 1000, 5000, 10000))

prediction <- predict(model, new_data) %>%
  as.data.frame() %>%
  cbind(new_data, .)

prediction

##################################################### CLASSIFICATION ################################################

CLASSI <- read.csv("classification.csv", sep=",",header = TRUE)

#train <- CLASSI %>% sample_frac(0.6)
#test <- anti_join(CLASSI, train)

test <- read.csv("test_classi.csv", sep=",",header = TRUE)

library(rpart)
library(caret)
library(e1071)

#+ v2ProductCategory
#+ v2ProductCategory

tree <- rpart(Transactions ~ operatingSystem + visits + medium ,
              data = CLASSI, method = "class", control=rpart.control(minsplit=0,cp=0))

test$prediction <- predict(tree, test, type = "class")
conf <- confusionMatrix(data = factor(test$prediction), reference = factor(test$Transactions))

#sensibilit? est la capacit? du mod?le ? pr?dire un positif quand la donn?e est r?ellement positive,
#sp?cificit? est la capacit? du mod?le ? pr?dire un n?gatif lorsqu'il y a vraiment un n?gatif.

conf$byClass["Sensitivity"]
conf$byClass["Specificity"]


library(rpart.plot) # Pour la repr?sentation de l'arbre de d?cision

prp(tree, extra = 1)

#Elagage de l'arbre avec le cp optimal
tree_Opt <- prune(tree,cp=tree$cptable[which.min(tree$cptable[,4]),1])

#Repr?sentation graphique de l'arbre optimal
prp(tree_Opt, extra = 1)


################################################### K-MEANS ########################################################

DATA <- read.csv("test.csv", sep=",",header = TRUE)

test_kmeans <- DATA[,2:7]

id <- DATA[,1]

rownames(test_kmeans) <- id

ratio_ss <- data.frame(cluster = seq(from = 1, to = 10, by = 1))
for (k in 1:10) {
  km_model <- kmeans(test_kmeans, k, nstart = 1000)
  ratio_ss$ratio[k] <- km_model$tot.withinss / km_model$totss
}

ggplot(ratio_ss, aes(cluster, ratio)) +  geom_line() + geom_point()

k_m <- kmeans (test_kmeans, 3, nstart = 1000)

id_cluster <- data.frame(k_m$cluster)
cluster_center <- data.frame(k_m$centers)
cluster_size <- data.frame(k_m$size)

test_kmeans$cluster <- k_m$cluster

ggplot(test_kmeans, aes(Transactions,Revenue, col = factor(cluster)))  + geom_point(size = 2, alpha = 0.8, position = "jitter")

k_m <- kmeans (test_kmeans, 3, nstart = 1000)

id_cluster <- data.frame(k_m$cluster)
cluster_center <- data.frame(k_m$centers)
cluster_size <- data.frame(k_m$size)

test_kmeans$cluster <- k_m$cluster

DATA$cluster <- k_m$cluster

write.csv(DATA,"data.csv", row.names = FALSE)

ggplot(test_kmeans, aes(Transactions,Revenue, col = factor(cluster)))  + geom_point(size = 2, alpha = 0.8, position = "jitter")

ggplot(test_kmeans, aes(x = Transactions)) + geom_histogram() + facet_wrap(~cluster)

ggplot(DATA, aes(x = Revenue, fill=cluster)) + geom_bar()

############################################## RANDOM FOREST #######################################################


FOREST <- read.csv("ML.csv", sep=",",header = TRUE)

install.packages("rpart.plot")

library(rpart)# Pour l'arbre de d?cision
library(rpart.plot) # Pour la repr?sentation de l'arbre de d?cision

#Cr?ation d'un dataset d'apprentissage et d'un dataset de validation
train <- CLASSI %>% sample_frac(0.8)
test <- anti_join(CLASSI, train)

#Construction de l'arbre
forest.Tree <- rpart(Transactions_2 ~ Device + medium +
                                      campaign + Page_views,
                                      data=train,method='class',
                                      control=rpart.control(minsplit=1,cp=0))

############################################# RANDOM FOREST #######################################################

lycee <- read.csv2("https://www.data.gouv.fr/s/resources/indicateurs-de-resultat-des-lycees-denseignement-general-et-technologique/20160401-163749/MEN-DEPP-indicateurs-de-resultats-des-LEGT-2015.csv",
                   sep = ";", header = TRUE, fileEncoding = "ISO-8859-15", na.strings = c(" ",
                                                                                          "."))