Feature Dominance and Explanation Stability in Imbalanced Purchase Intention Prediction

DOI

10.2991/978-94-6239-801-6_148How to use a DOI?

Keywords

Purchase intention; Class imbalance; Feature dominance; SHAP; E-commerce analytics

Abstract

Purchase intention prediction supports conversion management in e-commerce, but evaluation can be misleading when purchase sessions are imbalanced and conversion-related features dominate prediction. This study examines session-level purchase intention prediction using XGBoost, four imbalance-handling strategies, PageValues ablation, and SHAP-based explanation stability analysis. The Online Shoppers Purchasing Intention Dataset is used as the empirical setting. Among all evaluated models, the complete-feature XGBoost model exhibited the best overall ranking performance, yielding an F1 score of 0.6785, an area under the receiver operating characteristic curve of 0.9330, an area under the precision-recall curve of 0.7446, and a Matthews correlation coefficient of 0.6183. Class weighting and random oversampling did not provide consistent improvement, while SMOTE reduced several metrics. Removing PageValues and its zero-value indicator reduced F1-score to 0.4242 and PR-AUC to 0.3703, showing that performance depended heavily on conversion-derived information. SHAP results further show that SMOTE changed top-ranked explanations more than class weighting and random oversampling. Purchase intention models should therefore be judged by predictive performance, real-time feature availability, and explanation stability in both offline experiments and practical deployment.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Yiming Ding
PY  - 2026
DA  - 2026/10/23
TI  - Feature Dominance and Explanation Stability in Imbalanced Purchase Intention Prediction
BT  - Proceedings of the 2026 2nd International Conference on Financial Innovation and Marketing Management (FIMM 2026)
PB  - Atlantis Press
SP  - 1415
EP  - 1422
SN  - 2352-5428
UR  - https://doi.org/10.2991/978-94-6239-801-6_148
DO  - 10.2991/978-94-6239-801-6_148
ID  - Ding2026
ER  -