DEVELOPMENT OF MACHINE LEARNING TECHNIQUES FOR DETERMINING OPTIMAL PRICING AND INVENTORY POLICY WITH ORDER CANCELLATION UNDER THE CASH-ON-DELIVERY PAYMENT SCHEME.

Abstract

This study aims to analyze the optimal pricing and inventory policy of sellers in the context of order cancellations under the cash-on-delivery payment scheme. Additionally, it seeks to investigate the potential order cancellation behavior of customers in the same payment scheme. To achieve these objectives, the study employs machine learning models (ML models), which offer novel insights compared to previous research that primarily relies on periodic review systems and lacks empirical data-driven analysis. The collection of primary data involved the distribution of 315 questionnaires through Google Docs, utilizing a purposive sampling approach. These surveys were targeted at those who engage in online ordering within the United Kingdom. The evaluation of model accuracy for three machine learning predictive models was conducted, revealing that the logistic regression model exhibited the highest prediction accuracy of 84%, accompanied by a precision rate of 69%. The findings from the logistic regression analysis indicate that the potential cancellation of customer orders has a statistically significant negative effect on the seller's optimal pricing and inventory policy. This implies that the implementation of optimal pricing and inventory policies by the seller is associated with a decrease in the likelihood of customers canceling their orders. Hence, it is imperative for the vendor or seller of online products to implement a sustainable and optimal price and inventory policy in order to mitigate client order cancellations within the cash-on-delivery framework.