Online news feed data mining and prediction

Abstract

Data mining and prediction systems have been the center of attraction since information retrieval came into existence. Most IT companies spend a lot of resources on such analysis and systems to improve their performance and generate more revenue depending on the nature of work that they do. Online News Feed Prediction System aims to provide an analysis and comparison of various prediction techniques by using different methods of implementation. UCI repository contains a collection of databases pertaining to different topics. News popularity in multiple social media is one such dataset containing information about news topics from different sources, sentiment analysis of title and headline, topic that they are related to, publishing date, popularity score in various social media platforms. Python, R and Weka have been used on this data set to implement data preprocessing, visualization and prediction techniques like Random Forest, Decision Tree and SVM. Moreover, there is dataset on the analysis of the score for every twenty minutes for the social media platforms chosen. Analysis on these platforms helps in developing a system to reach a wider audience. News agencies can use this system to increase their profit and visibility. This paper aims to realize the ways to obtain these results.