Overview
Can you use this data set to make an algorithm able to determine if an article is fake news or not ?
Data Description
Fake.csv file contains a list of articles considered as “fake” news. True.csv contains a list of articles considered as “real” news. Both the files contain
- The title of the article
- The text of the article
- The subject of the article
- The date that this article was posted at
You can find the dataset here.
Files
- Fake.csv
- True.csv
So let’s begin here…
import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
from nltk.corpus import stopwords
from nltk.stem.porter import PorterStemmer
from string import punctuation
import keras
from keras.models import Sequential
from keras.layers import Dense
from sklearn.metrics import classification_report,confusion_matrix,accuracy_score
from sklearn.model_selection import train_test_split
from sklearn.feature_extraction.text import CountVectorizer
import os
for dirname, _, filenames in os.walk('/kaggle/input'):
for filename in filenames:
print(os.path.join(dirname, filename))
Using TensorFlow backend.
/kaggle/input/fake-and-real-news-dataset/Fake.csv
/kaggle/input/fake-and-real-news-dataset/True.csv
Load Data
real = pd.read_csv("../input/fake-and-real-news-dataset/True.csv")
fake = pd.read_csv("../input/fake-and-real-news-dataset/Fake.csv")
real.head()
| title | text | subject | date | |
|---|---|---|---|---|
| 0 | As U.S. budget fight looms, Republicans flip t... | WASHINGTON (Reuters) - The head of a conservat... | politicsNews | December 31, 2017 |
| 1 | U.S. military to accept transgender recruits o... | WASHINGTON (Reuters) - Transgender people will... | politicsNews | December 29, 2017 |
| 2 | Senior U.S. Republican senator: 'Let Mr. Muell... | WASHINGTON (Reuters) - The special counsel inv... | politicsNews | December 31, 2017 |
| 3 | FBI Russia probe helped by Australian diplomat... | WASHINGTON (Reuters) - Trump campaign adviser ... | politicsNews | December 30, 2017 |
| 4 | Trump wants Postal Service to charge 'much mor... | SEATTLE/WASHINGTON (Reuters) - President Donal... | politicsNews | December 29, 2017 |
fake.head()
| title | text | subject | date | |
|---|---|---|---|---|
| 0 | Donald Trump Sends Out Embarrassing New Year’... | Donald Trump just couldn t wish all Americans ... | News | December 31, 2017 |
| 1 | Drunk Bragging Trump Staffer Started Russian ... | House Intelligence Committee Chairman Devin Nu... | News | December 31, 2017 |
| 2 | Sheriff David Clarke Becomes An Internet Joke... | On Friday, it was revealed that former Milwauk... | News | December 30, 2017 |
| 3 | Trump Is So Obsessed He Even Has Obama’s Name... | On Christmas day, Donald Trump announced that ... | News | December 29, 2017 |
| 4 | Pope Francis Just Called Out Donald Trump Dur... | Pope Francis used his annual Christmas Day mes... | News | December 25, 2017 |
Pre-process Data
Let’s add another column in data set as ‘category’ where category will be 1 if news is real and 0 if news is fake.
real['category']=1
fake['category']=0
Now we will concatenate both the datasets
df = pd.concat([real,fake])
df.head()
| title | text | subject | date | category | |
|---|---|---|---|---|---|
| 0 | As U.S. budget fight looms, Republicans flip t... | WASHINGTON (Reuters) - The head of a conservat... | politicsNews | December 31, 2017 | 1 |
| 1 | U.S. military to accept transgender recruits o... | WASHINGTON (Reuters) - Transgender people will... | politicsNews | December 29, 2017 | 1 |
| 2 | Senior U.S. Republican senator: 'Let Mr. Muell... | WASHINGTON (Reuters) - The special counsel inv... | politicsNews | December 31, 2017 | 1 |
| 3 | FBI Russia probe helped by Australian diplomat... | WASHINGTON (Reuters) - Trump campaign adviser ... | politicsNews | December 30, 2017 | 1 |
| 4 | Trump wants Postal Service to charge 'much mor... | SEATTLE/WASHINGTON (Reuters) - President Donal... | politicsNews | December 29, 2017 | 1 |
Check for NULL values
df.isna().sum()
title 0
text 0
subject 0
date 0
category 0
dtype: int64
Total number of news
df.title.count()
44898
Number of news grouped by Subject
df.subject.value_counts()
politicsNews 11272
worldnews 10145
News 9050
politics 6841
left-news 4459
Government News 1570
US_News 783
Middle-east 778
Name: subject, dtype: int64
We now concatenate Text, Title and Subject in Text.
df['text'] = df['text'] + " " + df['title'] + " " + df['subject']
del df['title']
del df['subject']
del df['date']
df.head()
| text | category | |
|---|---|---|
| 0 | WASHINGTON (Reuters) - The head of a conservat... | 1 |
| 1 | WASHINGTON (Reuters) - Transgender people will... | 1 |
| 2 | WASHINGTON (Reuters) - The special counsel inv... | 1 |
| 3 | WASHINGTON (Reuters) - Trump campaign adviser ... | 1 |
| 4 | SEATTLE/WASHINGTON (Reuters) - President Donal... | 1 |
stop = set(stopwords.words('english'))
pnc = list(punctuation)
stop.update(pnc)
stemmer = PorterStemmer()
def stem_text(text):
final_text = []
for i in text.split():
if i.strip().lower() not in stop:
word = stemmer.stem(i.strip())
final_text.append(word)
return " ".join(final_text)
df.text = df.text.apply(stem_text)
Splitting dataset in train set and test set
X_train,X_test,y_train,y_test = train_test_split(df.text,df.category)
cv = CountVectorizer(min_df=0,max_df=1,ngram_range=(1,2))
cv_train = cv.fit_transform(X_train)
cv_test = cv.transform(X_test)
print('BOW_cv_train:',cv_train.shape)
print('BOW_cv_test:',cv_test.shape)
BOW_cv_train: (33673, 1950850)
BOW_cv_test: (11225, 1950850)
Defining Model
model = Sequential()
model.add(Dense(units = 100 , activation = 'relu' , input_dim = cv_train.shape[1]))
model.add(Dense(units = 50 , activation = 'relu'))
model.add(Dense(units = 25 , activation = 'relu'))
model.add(Dense(units = 10 , activation = 'relu'))
model.add(Dense(units = 1 , activation = 'sigmoid'))
Compile Model
model.compile(optimizer = 'adam' , loss = 'binary_crossentropy', metrics = ['accuracy'])
Fit Model
model.fit(cv_train,y_train , epochs = 5)
Epoch 1/5
33673/33673 [==============================] - 542s 16ms/step - loss: 0.6904 - accuracy: 0.5224
Epoch 2/5
33673/33673 [==============================] - 559s 17ms/step - loss: 0.1654 - accuracy: 0.9452
Epoch 3/5
33673/33673 [==============================] - 571s 17ms/step - loss: 0.0428 - accuracy: 0.9885
Epoch 4/5
33673/33673 [==============================] - 575s 17ms/step - loss: 0.0414 - accuracy: 0.9888
Epoch 5/5
33673/33673 [==============================] - 568s 17ms/step - loss: 0.0418 - accuracy: 0.9888
Prediction
pred = model.predict(cv_test)
for i in range(len(pred)):
if(pred[i] > 0.5):
pred[i] = 1
else:
pred[i] = 0
Evaluate Model
Accuracy
accuracy_score(pred,y_test)
0.90271714922049
Classification Report
cv_report = classification_report(y_test,pred,target_names = ['0','1'])
print(cv_report)
precision recall f1-score support
0 0.92 0.89 0.91 5929
1 0.88 0.91 0.90 5296
accuracy 0.90 11225
macro avg 0.90 0.90 0.90 11225
weighted avg 0.90 0.90 0.90 11225
Confusion Matrix
cm_cv = confusion_matrix(y_test,pred)
cm_cv = pd.DataFrame(cm_cv, index=[0,1], columns=[0,1])
cm_cv.index.name = 'Actual'
cm_cv.columns.name = 'Predicted'
plt.figure(figsize = (10,10))
sns.heatmap(cm_cv,cmap= "Blues",annot = True, fmt='')
