An ensemble classification approach for prediction of user’s next location based on Twitter data

Abstract

The study presents a machine learning approach to predict a user’s next location of visit based on their previous tweets and personality traits. The future behavior of the people is reflected by their profile and past behaviors; this motivates the work presented in this study. Nowadays, people are fond of sharing their experiences related to travel, a visit to a restaurants or hotel or some historic places etc. on the social media platforms. Twitter, Instagram and Facebook are one of the popular social media platforms where this sharing of information can be found. This study used Twitter data for the analysis and to develop a model using machine learning (ML) to predict a user’s next visiting location. ML enables a computer to learn from the historical records and use this knowledge for prediction and decision making for new data. Prediction accuracy is one of the important and required parameter for any prediction model. If prediction model does not provide good prediction accuracy then the model cannot be accepted as reliable. Considering this phenomena, this study used an ensemble classification approach (ESA) to develop a prediction model for the problem under study. ESA trains different classifiers on the same data and use voting method to select the most accurate prediction. This study used ESA for first time to perform such type of study and results reveal that ESA certainly enhances the prediction accuracy of the model which is highly desirable.