Optimizing Discount Strategies Using Machine Learning / Deep Learning Models with RPA Tool

Abstract

Optimizing discount strategies using Machine Learning (ML) and Deep Learning (DL) models helps maximize sales while maintaining profitability and optimal discount rates. Unplanned discount strategies may not always yield the best results. In this study, a dataset was collected from Amazon using UiPath, a Robotic Process Automation (RPA) tool. The dataset includes television product details such as price, discount percentage, ratings, and other factors. Among ML models, Random Forest performed the best, achieving a <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$10-12 {\%}$</tex> improvement over Linear Regression and Decision Trees in terms of accuracy due to its ability to capture complex relationships in data. XGBoost showed slightly better performance than Random Forest, providing a <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$3-5 {\%}$</tex> improvement due to its gradient boosting mechanism. In DL models, the Transformer-based model outperformed MLP and LSTM, reducing the error by 15-20% compared to MLP and by 8-10% compared to LSTM. The Transformer model effectively captured dependencies in the dataset, making it the most accurate for predicting optimal discounts. The findings provide insights into the most effective pricing strategies that businesses can use to maximize revenue while maintaining profitability. The final output of this research is a data-driven recommendation system that suggests optimal discount rates for products to enhance sales while minimizing revenue loss.