Correlation Enhanced Machine Learning Approach based Online News Popularity Prediction

Abstract

News popularity is the maximum growth of attention given for particular news article. The popularity of online news depends on various factors such as the number of social media, the number of visitor comments, the number of Likes, etc. It is therefore necessary to build an automatic decision support system to predict the popularity of the news as it will help in business intelligence too. The work presented in this study aims to find the best model to predict the popularity of online news using machine learning methods. Initially, correlation techniques are used to gain dependence on the popularity received from an article and to obtain attributes or characteristics that are optimal for subsequent classification. Data has been procured from UCI Machine Learning Repository with 39644 articles with sixty condition attributes and one decision attribute. Then different learning algorithms such as Proposed Hybrid SVM-RF, AdaBoost, LPBoost, and KNN are implemented in order to predict the news popularity. The performance of system is tested on the dataset which comes from UCI machine learning repository. The prediction performances of all methodologies are studied by considering evaluation measures. Hybrid SVM-RF turns out to be the best model for prediction and it has achieved accuracy of 99.6% for binary classification. Further this work is enhanced for multiclass classification with different learning algorithms such as Proposed Hybrid SVM-RF, Naïve Bayes and KNN. Hybrid SVM-RF had achieved the accuracy of about 73% accuracy as compared with other classifiers.