Improved Accuracy for Fake News in Social Media Using Logistic Regression Comparing Naive Bayes Classifier

Abstract

The fundamental purpose of the work is to detect the Fake News in Social Media with the use of Machine Learning Algorithms.TRUE and FAKE dataset is used to detect false news. This dataset carries the record of data i.e. TRUE or FAKE news. Fake News detection is accomplished via Logistic Regression and Naive Bayes classifier. Naive-Bayes algorithm is a simple approach mainly used for classification. Sample size has been determined to be 20 for both the groups using G Power 80%. Logistic Regression algorithm provides mean accuracy of 97.5% when compared to Naive Bayes algorithm with mean accuracy of 89.43%. Statistical significance value is obtained as 0.002 (p<0.05). Logistic Regression has extensively higher accuracy than Naive Bayes algorithm.