Forecasting E-Commerce User Engagement: An Analysis Comparing Machine Learning and Deep Learning Models on Session Behavior

In this new internet era, a sizable percentage of purchases have begun to be occur online rather than in real life. This revolution has caused a whole new branch in economics called e commerce where new parameters were set when it comes to selling products. The use of online shopping has generally given convenience for buy where they can now purchase products in the comfort of their home rather than take a long visit to purchase something you are not even sure is available or nearly as good as marketed. It also allows easy access to reviews which help assure consumers of false products and scams. The most important part however, is that it also allows a great convenience for buyers in all places of the world to sell products and make money. In order to help forecast the tolerance of customers and their behavior, we propose using “machine learning algorithms such as XGBoost, Random Forest, Logistic Regression and CatBoost”; and deep learning (Convolutional Neural Network)algorithms such as “Gated Recurrent Unit, Long Short-Term Memory, and Temporal Convolutional Model (or more often TCN - Temporal Convolutional Network)” in order to help predict the likelihood of a customer purchasing or viewing a product based on the type, data, brand, and price of the product. The achieved results are measured in terms of “Accuracy (ACC), Precision (PREC), Recall (REC), and F1-score (F1)”. We believe these results will maximize profits in online shopping and help navigate consumer behavior properly, pursue economic development, grow revenue generation and improve sales prediction.