Predictive Analytics for Reverse Logistics: Forecasting Return Volumes in E-Commerce

Abstract

The returns increase the logistics, warehousing, and customer service strain besides increasing the operating costs. It explores the potential of prediction analytics to improve the efficiency of reverse logistics, based on the case Study of Call Courier, in the Pakistani e-commerce industry. The project aims at knowing and predicting the trends of the product returns in order to aid in enhanced operation planning, cost control and customer satisfaction. First Chapter indicates high rate of expansion of the e-commerce and the ensuing problems to deal with products returns. The role of the reverse logistics, which deals with the processing of the returned goods, is now the very important part of the supply chain management because the returns rates directly influence the profitability and the use of resources. The issue this research will seek to solve is the absence of proper forecasting models that could forecast product returns and their business impacts on the logistics companies such as Call Courier. Quantitative research methodology is adopted, which is based on a ten-year synthetic dataset of the shipment operations of Call Courier. The variables in the data were the shipment identifiers, geographic characteristics, cash-on-delivery (COD) behavior, cost measures, and the return flags. The dataset was subjected to data preprocessing and feature engineering to clean, standardize the dataset and add time-based and distance-based features. The descriptive analytics indicated that the rates of returns were more in non-metro and remote locations, which was attributed primarily to failures in delivering and COD non-collection but in metro regions, there was higher success in delivery.