Analyzing Online Shopping Behavior: A Machine Learning Approach to Revenue Prediction, Eimad Abusham, Salahaldin Abdulkader
Eimad Abusham1, and Salahaldin Abdulkader 2
1,2 Faculty of Computing and IT, Sohar University, PO Box 44, Sohar, PC 311, Oman; eabusham@su.edu.om,
2 Faculty of Business, Sohar University, PO Box 44, Sohar, PC 311, Oman; SAbdulkader@su.edu.om
Abstract:
This competition has come hand in hand with the fast-growing e-commerce business, and there’s a need to develop good Models for revenue predictions so as to manage marketing plans well. This paper aims at investigating the use of various machine learning algorithms such as Logistic Regression, Random Forest and Gradient Boosting needed in the prediction of the intentions of the online shoppers in relation to revenue. In this case, we use the Online Shoppers Intention dataset obtained from the UCI Machine Learning Repository to understand key drivers affecting buying behaviors. The evaluation criterion used include accuracy and ROC, based on which it was established that Gradient Boosting has the highest ROC of 0.930 while Random Forest has the highest accuracy of 0.893 more than that of the Logistic Regression with accuracy of 0.873, ROC of 0.891. These observations show that integrating multiple weak models provides higher predictive accuracy and generalizability when modeling irregularity of shoppers’ behavior. The results of this study are useful in guiding the improvement of targeted advertising and customer interactions specifically in the field of online retail while also advancing the more extensive area of e-commerce analytics.