| dc.description.abstract | Timely order fulfillment is a critical determinant of customer satisfaction and operational efficiency in e-commerce logistics, particularly in emerging economies such as Bangladesh. This study examines on-time delivery performance using Daraz import shipment data, representing one of the largest cross-border e-commerce logistics operations in the country. The research develops a machine learning-based predictive framework to identify operational and customer-related factors contributing to delivery delays. Supervised classification models, including tree-based algorithms, are applied to a large shipment dataset encompassing warehouse allocation, shipment mode, product characteristics, customer interactions, and promotional attributes. Model performance is assessed using standard classification metrics to ensure robustness and reliability. The results reveal that shipment mode, product weight, warehouse block, customer care calls, and discount levels significantly influence delivery timeliness. The findings provide actionable insights for e-commerce platforms and logistics managers, offering data-driven recommendations to improve import shipment reliability and enhance last-mile delivery performance in Bangladesh. | en_US |
