Citation:

Md Rasel Uddin. "Decoupling the Nature of the Customer: A Quantile Regression Approach to Heterogeneous Engagement Dynamics" International Research Journal of Economics and Management Studies, Vol. 5, No. 4, pp. 208-214, 2026. Crossref. http://doi.org/10.56472/25835238/IRJEMS-V5I4P125

Abstract:

Global e-commerce growth has turned marketing into a complex, multi-channel game. This complexity makes it difficult for businesses to attribute revenue to the right sources. With so much data noise, clear attribution is now a major challenge. Traditional regression models often break down when marketing and advertising costs overlap too heavily. When the Variance Inflation Factor (VIF) climbs above 20.0, standard coefficient estimates essentially become useless. This study addresses these flaws by developing a high-fidelity framework. Our approach isolates the actual drivers of sales revenue and customer satisfaction, providing a reliable alternative to standard analytical methods. To ensure the preservation of multivariate integrity under the Missing at Random (MAR) assumption, the framework implements MICE, which stands for Multivariate Imputation by Chained Equations. We address the issues of redundant dimensionality and heteroscedasticity by engineering a synthetic feature, Capital Force (Cf), through Principal Component Analysis. Instead of relying on standard linear methods, our predictive framework uses an Extreme Gradient Boosting (XGBoost) regressor. We then apply SHAP values to pull back the "black box" of the model for better transparency. Finally, to see how marketing sensitivity shifts across economic tiers, we employed Quantile Regression at the 10th, 50th, and 95th percentiles. Our model achieved an R^2 of 0.84, but the real story lies in the sharp split it revealed in consumer behavior. At the lower end of the spectrum, price incentives are the main driver. However, as you move into high-tier segments, that influence fades; instead, these customers are moved by how recently and consistently they engage with digital platforms. These findings underscore a necessary strategic pivot from uniform marketing models to bifurcated approaches that prioritize relational touchpoints for high-value retention. Ultimately, a persistent digital presence emerges as a critical hedge against satisfaction decay, ensuring long-term loyalty in volatile e-commerce markets.

References:

[1] Ayestarán, R. (2021). Customer Relationship Marketing (CRM) for the Design of Strategies in Digital Marketing. Environment and Innovation Strategies to Promote Growth and Sustainability, 108–119. https://doi.org/10.1201/9781003136712-7
[2] Cai, K., & Rodavia, M. R. (2023). XGBoost Analysis based on Consumer Behavior. Frontiers in Computing and Intelligent Systems, 5(2), 85–89. https://doi.org/10.54097/FCIS.V5I2.12974
[3] Chapter 10: Predictive approaches to customer loyalty: the impact of missing data on the predictability of customer loyalty models in: Handbook of Research on Customer Loyalty. (n.d.). Retrieved April 28, 2026, from https://www.elgaronline.com/edcollchap/book/9781800371637/book-part-9781800371637-16.xml
[4] Curnow, E., Carpenter, J. R., Heron, J. E., Cornish, R. P., Rach, S., Didelez, V., Langeheine, M., & Tilling, K. (2023). Multiple imputation of missing data under missing at random: compatible imputation models are not sufficient to avoid bias if they are mis-specified. Journal of Clinical Epidemiology, 160, 100–109. https://doi.org/10.1016/j.jclinepi.2023.06.011
[5] Dhoke, H., Nayak, S., & Khilar, P. M. (2025). Data Imputation using Machine Learning for Enhanced Predictive Modeling. 1–5. https://doi.org/10.1109/AICDMB64359.2025.11277656
[6] E-COMMERCE DATA ANALYSIS – IJSREM. (n.d.). Retrieved April 28, 2026, from https://ijsrem.com/download/e-commerce-data-analysis/
[7] Hanif, I. (2020). Implementing Extreme Gradient Boosting (XGBoost) Classifier to Improve Customer Churn Prediction. https://doi.org/10.4108/EAI.2-8-2019.2290338
[8] Ho, T., Nguyen, S., Nguyen, H., Nguyen, N., Man, D. S., & Le, T. G. (2023). An Extended RFM Model for Customer Behaviour and Demographic Analysis in Retail Industry. Business Systems Research Journal, 14(1), 26–53. https://doi.org/10.2478/BSRJ-2023-0002
[9] Jolani, S., & van Buuren, S. (2024). Random Indicator Imputation for Missing Not At Random Data. https://arxiv.org/pdf/2404.14534
[10] Kalnins, A. T., & Praitis Hill, K. (2023). The Myths of the VIF: Why Variance Inflation Factors are Deceptive as Multicollinearity Diagnostics. Https://Doi.Org/10.5465/AMPROC.2023.19324abstract, 2023(1). https://doi.org/10.5465/AMPROC.2023.19324ABSTRACT
[11] Kaur, S., & Kapil, Dr. R. (2023). Measuring Customer Engagement in Digital Marketing. Journal for ReAttach Therapy and Developmental Diversities, 6(6s), 984–992. https://doi.org/10.53555/JRTDD.V6I6S.2951
[12] Kiradoo, D. G. (2023). Predictive Analytics in E-Commerce: A Quantitative Approach to Optimizing Customer Experience. Predictive Analytics in E-Commerce: A Quantitative Approach to Optimizing Customer Experience. https://doi.org/10.59646/ECEBOOKC28/007
[13] Lin, J., & Fukuyama, J. (2026). A comparative analysis of machine learning models in SHAP analysis. https://arxiv.org/pdf/2604.07258
[14] Liu, J., Wang, Y., & Lin, H. (2025). Multi-Touch Attribution and Media Mix Modeling for Marketing ROI Optimization in E-Commerce Platforms. Frontiers in Business and Finance, 2(02), 378–398. https://doi.org/10.71465/FBF528
[15] Malhotra, N. K., & Agarwal, J. (2021). Customer Relationship Marketing: Analytical Models and Strategies. Customer Relationship Marketing, 165–196. https://doi.org/10.1142/9781944659721_0006
[16] Nakayama, Y., & Araki, N. (2020). An Exploratory Study on the Clumpiness Measure of Intertransaction Times: How Is It Useful for Customer Relationship Management? Studies in Classification, Data Analysis, and Knowledge Organization, 329–339. https://doi.org/10.1007/978-981-15-3311-2_26
[17] Rao, T., Agarwal, S., & Singh, N. (2023). An Empirical Evaluation of Shapley Additive Explanations: A Military Implication. 2023 10th IEEE Uttar Pradesh Section International Conference on Electrical, Electronics and Computer Engineering, UPCON 2023, 1390–1397. https://doi.org/10.1109/UPCON59197.2023.10434608
[18] Sanni, J. O., Atima, M. E., & Attah, A. (2022). Systematic Review of Attribution Modeling Methods Resolving Bias in Multi Touch Journeys. Shodhshauryam International Scientific Refereed Research Journal, 5(1), 304–334. https://doi.org/10.32628/SHISRRJ247135
[19] Shadab, M., & Faisal, M. (2025). IMPLEMENTATION AND EVALUATION OF CUSTOMER CHURN PREDICTION MODELS FOR E-BUSINESS SERVICES USING XGBOOST. International Journal of Applied Mathematics, 38(11s), 36–104. https://doi.org/10.12732/IJAM.V38I11S.1147
[20] Sim, D. Y. Y., & Wei, Z. (2023). XGBoost Regression Algorithms for Efficient Predictions on Inventory Sales and Management. Proceedings - 2023 6th International Conference on Electronics and Electrical Engineering Technology, EEET 2023, 66–71. https://doi.org/10.1109/EEET61723.2023.00033
[21] Soundarapandian, D. (2024). Algorithmic Framework for Retail Media Optimization and Consumer Engagement Enhancement. Journal of Business Intelligence and Data Analytics, 1(3), 1–7. https://doi.org/10.55124/jbid.v1i3.259
[22] Stella Nwabekee, U., Yetunde Abdul-Azeez, O., Ebele Agu, E., & Ignatius Ijomah, T. (2024). Digital transformation in marketing strategies: The role of data analytics and CRM tools. https://doi.org/10.56355/ijfrst.2024.3.2.0047
[23] Taşabat, S. E., Özçay, T., Sertbaş, S., & Akca, E. (2023). A New RFM Model Approach: RFMS. Industry 4.0 and the Digital Transformation of International Business, 143–172. https://doi.org/10.1007/978-981-19-7880-7_9/SAVE-RESEARCH
[24] Verma, A. (2025). A Comparative Study of Ad Attribution Models: Evaluating the Impact on ROI Measurement. European Journal of Computer Science and Information Technology, 13(35), 56–75. https://doi.org/10.37745/EJCSIT.2013/VOL13N355675

Keywords:

Capital Force, Customer Satisfaction, Digital Engagement, E-commerce, Principal Component Analysis, Quantile Regression, SHAP, XGBoost.