Behavioural Insights into Online Shoppers’ Purchase Intention Using Machine Learning Models

Abstract

With evolving dynamics in e-commerce, user behavior analysis on e-commerce websites has grown more crucial in customer experience improvement and conversion rate optimization. Predictive analytics is instrumental in deriving hidden patterns from user behavior and enabling data-informed decisions. However, amidst vast amounts of web traffic data, the majority of online retailers struggle to identify actionable behavioral cues that accurately predict buying intent on a consistent basis. This study addresses the issue of accurately predicting purchasing intention based on session-based user behavior and demographic data. By examining the "Online Shoppers Purchasing Intention" dataset available at the UCI Machine Learning Repository, this project aims to predict whether or not a user will make a purchase during a session. Using Python as the primary tool, the study employs data preprocessing, exploratory data analysis, feature selection, and machine learning algorithms like Logistic Regression, Random Forest, and Support Vector Machines. The performance of these algorithms is evaluated using accuracy, precision, recall, and F1-score. Preliminary results show that page value, bounce rate, and visit month have a significant influence on purchase likelihood. The results highlight the importance of behavioral data in predicting e-commerce outcomes. The results can be utilized to inform strategic planning in UX design, online marketing, and inventory management.