Predicting popularity of online articles using Random Forest regression

Abstract

Predictive analysis using machine learning has been gaining popularity in recent times. In this paper, the Random Forest regression model is used to predict popularity of articles from the Online News Popularity data set. The performance of the Random Forest model is investigated and compared with other models. Impact of standardization, regularization, correlation, high bias/high variance and feature selection on the learning models are also studied. Results indicate that, the Random Forest approach predicts popular/unpopular articles with an accuracy of 88.8%.