Predicting Customer Revenue in E-commerce Using Machine Learning a Case Study of the Google Merchandise Store

This research paper explores the use of machine learning algorithms to predict customer revenue in e-commerce, using the Google Analytics Customer Revenue Prediction dataset as a case study. The dataset contains anonymized data from the Google Merchandise Store, an e-commerce site that sells Google-branded merchandise. We use the data to build and evaluate different machine learning models that predict the natural log of the revenue per customer for each session, based on various features such as demographic information, traffic source, and behavior on the website. Our findings suggest that machine learning algorithms like (LGBM Regressor) can effectively predict customer revenue in e-commerce, with root mean squared error (7.18e-11), Mean squared error(5.1e-21), R-squared(0.3260359), Mean Absolute Error(1.43e-11) and time performance(32 s). We also identify the key features that are most predictive of customer revenue, including visit number, total page views, total hits, hours, session ID, and day of the month. Overall, our research demonstrates the potential of machine learning in improving customer revenue prediction in e-commerce and provides insights for e-commerce businesses to optimize their marketing and sales strategies.