Fake News Detection on Social Media using Logistic Regression and Compare Textual Property with Decision Tree Algorithm

Key Points

• To evaluate the effectiveness of logistic regression and decision tree algorithms in detecting fake news on social media.
• Used machine learning algorithms logistic regression (LR) and decision tree (DT) for detection.
• Assessed models via 10 fold cross validation on 44,898 labeled news articles from the WELFake Kaggle dataset.
• Applied TF-IDF vectorization with unigram features and removed English stop words for feature extraction.
• Logistic regression achieved a mean accuracy of 96.52% ±0.41% with a precision of 96.10%, recall of 95.60%, and F1 score of 95.85%.
• Decision tree reported an accuracy of 88.34% ± 1.12%.
• Statistically significant difference was noted between logistic regression and decision tree results ($p<0.001$).