-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathsourse_file.R
233 lines (151 loc) · 10.5 KB
/
sourse_file.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
library(tidytext)
library(readr)
library(ggplot2)
library(dplyr)
library(tokenizers)
library(plotrix)
library(dplyr)
naiveBayes <- setRefClass("naiveBayes",
# here it would be wise to have some vars to store intermediate result
# frequency dict etc. Though pay attention to bag of wards!
fields = list(
credible_probs = "table",
fake_probs = "table",
credible_news = "data.frame",
fake_news = "data.frame",
p_credible = "table",
p_fake = "table"
),
methods = list(
tokens_conditional_prob = function(tokens) {
occurrences <- table(unlist(tokens)) + 1
return (occurrences / sum(occurrences))
},
# prepare your training data as X - bag of words for each of your
# messages and corresponding label for the message encoded as 0 or 1
# (binary classification task)
fit_data = function(dataf, splitted_stop_words)
{
dataf$text <- paste(dataf$Headline,dataf$Body, sep = " ")
dataf$text <- tokenize_words(dataf$text, lowercase = TRUE, strip_numeric = TRUE, strip_punct = TRUE, stopwords = splitted_stop_words)
credible_news <<- subset(dataf, Label == 'credible')
fake_news <<- subset(dataf, Label == 'fake')
credible_probs <<- tokens_conditional_prob(credible_news$text)
fake_probs <<- tokens_conditional_prob(fake_news$text)
truth_rate <<- credible_probs %>%
as.data.frame() %>%
arrange(desc(Freq))
fake_rate <<- fake_probs %>%
as.data.frame() %>%
arrange(desc(Freq))
truth_top <<- truth_rate[1:5, ]
truth_top$Fake <<- c(0, 0, 0, 0, 0)
for (j in 1:5){
for (i in 1:nrow(fake_rate)){
if (fake_rate[i,1] == truth_top[j,1]){
truth_top[j, 3] <<- fake_rate[i,2]
}
}
}
popular <- as.matrix(data.frame(will = c(truth_top[1, 2], truth_top[1, 3]),
can = c(truth_top[2, 2], truth_top[2, 3]),
one = c(truth_rate[3, 2], truth_top[3, 3]),
game = c(truth_top[4, 2], truth_top[4, 3]),
news = c(truth_rate[5, 2], truth_top[5, 3])))
barplot(popular, main = "Probability of top-words",
col = c("green", "red"),
beside = TRUE)
legend("topright",
legend = c("Credible", "Fake"),
fill = c("green", "red"))
p_credible <<- table(credible_news$Label) / sum(table(dataf$Label))
p_fake <<- table(fake_news$Label) / sum(table(dataf$Label))
return (dataf)
},
fit_test = function(dataf, splitted_stop_words)
{
dataf$text <- paste(dataf$Headline, dataf$Body, sep = " ")
dataf$text <- tokenize_words(dataf$text, lowercase = TRUE, strip_numeric = TRUE, strip_punct = TRUE, stopwords = splitted_stop_words)
return (dataf)
},
# return prediction for a single message
predict_for_dataf = function(dataf)
{
dataf$res <- NA
index = 1
for (news in dataf$text) {
dataf[index, 6] = predict(news)
index = index + 1
}
dataf <- dataf[c('Headline', 'Body', 'Label', 'res')]
return (dataf)
},
predict = function(news)
{
credible_guess = p_credible
fake_guess = p_fake
res = 1
for (word in unlist(news)) {
credible_word <- credible_probs[word]
fake_word <- fake_probs[word]
if (!is.na(credible_probs[word]) & !is.na(fake_probs[word])) {
credible_guess = (credible_probs[word] * p_credible) / (credible_probs[word] * p_credible + fake_probs[word] * p_fake)
}
if (!is.na(fake_probs[word]) & !is.na(credible_probs[word])) {
fake_guess = (fake_probs[word] * p_fake) / (credible_probs[word] * p_credible + fake_probs[word] * p_fake)
}
res = res * (credible_guess / fake_guess)
}
if (res > 1) {
return ('credible')
}
return ('fake')
},
# score you test set so to get the understanding how well you model
# works.
# look at f1 score or precision and recall
# visualize them
# try how well your model generalizes to real world data!
score = function(dataf)
{
tp = sum(table(which(dataf$Label == 'credible' & dataf$res == 'credible')))
fp = sum(table(which(dataf$Label == 'fake' & dataf$res == 'credible')))
fn = sum(table(which(dataf$Label == 'credible' & dataf$res == 'fake')))
precision = tp / (tp + fp)
recall = tp / (tp + fn)
f1score = 2 * (precision * recall) / (recall + precision)
return (f1score)
},
score_visual = function(dataf)
{
true_positive = table(dataf$res=="credible" & dataf$Label=="credible")["TRUE"]
true_negative = table(dataf$res=="fake" & dataf$Label=="fake")["TRUE"]
false_negative = table(dataf$res=="fake" & dataf$Label=="credible")["TRUE"]
false_positive = table(dataf$res=="credible" & dataf$Label=="fake")["TRUE"]
slices <- c(true_positive, true_negative, false_positive, false_negative)
lbls <- c("True Positive", "True Negative", "False Positive", "False Negative")
pct <- round(slices/sum(slices)*100)
lbls <- paste(lbls, pct)
lbls <- paste(lbls,"%",sep="")
pie(slices, labels = lbls, col=rainbow(length(lbls)), main="Prediction accuracy")
}
)
)
stop_words_path <- "/home/fazhur/R_Projects/Lab1/stop_words.txt"
test_path <- "/home/fazhur/R_Projects/Lab1/2-fake_news/test.csv"
train_path <- "/home/fazhur/R_Projects/Lab1/2-fake_news/train.csv"
main_f <- function(test_path, train_path, stop_words_path){
test <- read.csv(test_path, stringsAsFactors = FALSE)
data <- read.csv(train_path, stringsAsFactors = FALSE)
stop_words <- read_file(stop_words_path)
splitted_stop_words <- strsplit(stop_words, split='\n')[[1]]
model = naiveBayes()
data <- model$fit_data(data, splitted_stop_words)
test <- model$fit_test(test, splitted_stop_words)
test <- model$predict_for_dataf(test);
print('score - >')
print(model$score(test))
print(model$score_visual(test))
return (test)
}
test <- main_f(test_path, train_path, stop_words_path)