Abstract

The growth of e-commerce has made accurate demand forecasting and well-judged discount decisions central to pricing strategy. This study develops and evaluates a machine learning framework that forecasts order-level demand and uses the forecast to identify revenue-maximizing discount levels. The analysis uses a publicly available dataset of 30,600 Indian e-commerce orders spanning 36 months (January 2023 – December 2025), covering eight product categories and five geographic zones. Four regression models (Linear Regression, Ridge Regression, Random Forest, and XGBoost) were tuned with time-series cross-validation on the first 30 months, evaluated on the final six months, and compared with two simple baselines. XGBoost produced the most accurate demand forecast (MAE = 12.83 units, R² = 0.413), reducing error by 21.4% compared with a naive baseline, although its advantage over Linear Regression was small (1.7%). A revenue benchmark showed that the near-perfect accuracy obtained when units sold is used as an input (R² = 0.999) reflects the identity revenue = price × units rather than forecasting skill; without units sold, a two-stage demand × price forecast achieved R² = 0.663. SHAP analysis identified sales events, customer age, geographic zone, and discount depth as the main drivers of demand, while product price level, category, and calendar month had little effect. A discount simulation showed that revenue peaks at moderate discounts of roughly 20–45% and that deeper discounts consistently reduce revenue.

References

• K. J. Ferreira, B. H. A. Lee, and D. Simchi-Levi, “Analytics for an online retailer: Demand forecasting and price optimization,” Manufacturing & Service Operations Management, vol. 18, no. 1, pp. 69–88, 2016.
• T. Hastie, R. Tibshirani, and J. Friedman, The Elements of Statistical Learning: Data Mining, Inference, and Prediction, 2nd ed. New York, NY, USA: Springer, 2009.
• S. Makridakis, E. Spiliotis, and V. Assimakopoulos, “M5 accuracy competition: Results, findings, and conclusions,” International Journal of Forecasting, vol. 38, no. 4, pp. 1346–1364, 2022.
• L. Breiman, “Random forests,” Machine Learning, vol. 45, no. 1, pp. 5–32, 2001.
• T. Chen and C. Guestrin, “XGBoost: A scalable tree boosting system,” in Proc. 22nd ACM SIGKDD Int. Conf. Knowledge Discovery and Data Mining, San Francisco, CA, USA, 2016, pp. 785–794.
• S. M. Lundberg and S.-I. Lee, “A unified approach to interpreting model predictions,” in Advances in Neural Information Processing Systems 30, Long Beach, CA, USA, 2017, pp. 4765–4774.
• R. L. Phillips, Pricing and Revenue Optimization. Stanford, CA, USA: Stanford University Press, 2005.
• K. T. Talluri and G. J. van Ryzin, The Theory and Practice of Revenue Management. New York, NY, USA: Springer, 2004.
• C. Bergmeir and J. M. Benítez, “On the use of cross-validation for time series predictor evaluation,” Information Sciences, vol. 191, pp. 192–213, 2012.
• S. Shukla, “Indian E-commerce Pricing, Revenue Growth Dataset,” Kaggle. [Online]. Available: https://www.kaggle.com/datasets/shukla922/indian-e-commerce-pricing-revenue-growth. [Accessed: May 2026].
• F. Pedregosa et al., “Scikit-learn: Machine learning in Python,” Journal of Machine Learning Research, vol. 12, pp. 2825–2830, 2011.
• D. C. Montgomery, E. A. Peck, and G. G. Vining, Introduction to Linear Regression Analysis, 6th ed. Hoboken, NJ, USA: Wiley, 2021. A. E. Hoerl and R. W. Kennard, “Ridge regression: Biased estimation for nonorthogonal problems,” Technometrics, vol. 12, no. 1, pp. 55–67, 1970.