Conformal Predictive Distributions for Order Fulfillment Time Forecasting

TL;DR

This paper presents a new distributional forecasting framework for order fulfillment time in e-commerce, using conformal predictive systems and machine learning to improve accuracy and reliability over traditional rule-based methods.

Contribution

It introduces a novel, model-agnostic framework combining conformal prediction and spatiotemporal features for more accurate and reliable order fulfillment time forecasts.

Findings

Distributional forecasts are competitive and reliable.

Machine learning point predictions outperform rule-based systems.

Achieved up to 14% higher accuracy and 75% better late delivery detection.

Abstract

Accurate estimation of order fulfillment time is critical for e-commerce logistics, yet traditional rule-based approaches often fail to capture the inherent uncertainties in delivery operations. This paper introduces a novel framework for distributional forecasting of order fulfillment time, leveraging Conformal Predictive Systems and Cross Venn-Abers Predictors -- model-agnostic techniques that provide rigorous coverage or validity guarantees. The proposed machine learning methods integrate granular spatiotemporal features, capturing fulfillment location and carrier performance dynamics to enhance predictive accuracy. Additionally, a cost-sensitive decision rule is developed to convert probabilistic forecasts into reliable point predictions. Experimental evaluation on a large-scale industrial dataset demonstrates that the proposed methods generate competitive distributional forecasts, while machine learning-based point predictions significantly outperform the existing rule-based system -- achieving up to 14% higher prediction accuracy and up to 75% improvement in identifying late deliveries.