Uncovering App Activity Patterns Behind E-Commerce App Usage: An Explainable AI Framework

Myounggu Lee, College of Business Administration, Konkuk University, South KoreaFollow
Hye-Jin Kim, School of Business and Technology Management, College of Business, Korea Advanced Institute of Science and Technology, South KoreaFollow

ORCID

Myounggu Lee: https://orcid.org/0000-0001-8398-9165

Abstract

In today's competitive mobile ecosystem, understanding how general app activity influences e-commerce app usage is crucial for marketers. This study introduces an explainable AI (XAI) framework using SHapley Additive exPlanations (SHAP) to identify non-e-commerce apps associated with e-commerce app usage. Drawing on mobile app launch data from 1,000 Android users in South Korea, we train a machine learning model to predict usage and apply SHAP to quantify the marginal contribution of each app. Results show that apps such as YouTube, Naver, and KakaoTalk are strongly linked to e-commerce activity at the user level. By visualizing user–app associations and clustering users based on shared activity patterns, we uncover latent behavioral structures within the mobile ecosystem. This framework provides interpretable, user-level insights that extend attribution modeling and generate actionable implications for segmentation, personalized targeting, and app ecosystem design.

Recommended Citation

Lee, Myounggu and Kim, Hye-Jin (2026) "Uncovering App Activity Patterns Behind E-Commerce App Usage: An Explainable AI Framework," Asia Marketing Journal: Vol. 28 : Iss. 2 , Article 3.
Available at: https://doi.org/10.53728/2765-6500.1680