Multi-Target Decision Analytics for Logistics Delay Prediction and Operational Efficiency Enhancement

Abstract

Global logistics networks handle billions of shipments annually, yet studies indicate that over 20% of global freight experiences delays and inefficient routing contributes to nearly 15% higher operational costs. In traditional supply chains, delay detection and rerouting are managed manually or through static systems, often resulting in poor real-time decision-making and misallocation of resources. This work proposes a predictive analytics framework for smart logistics delay detection. The dataset consists of shipment records including delivery times, routes, resource allocation, and feasibility metrics. The preprocessing phase includes data cleaning, normalization, and transformation to prepare the dataset for multi-output predictive tasks. For baseline comparisons, existing models such as KNearest Neighbor with Classification and Regression Tree (KNN-CART) and Huber-CART are evaluated. The proposed method introduces a Decision Tree-based CART (DT-CART) that integrates three classifiers and one regression tree in a hybrid predictive architecture. Specifically, Classifier 1 detects rerouted shipments, Classifier 2 predicts resource allocation feasibility, Classifier 3 validates shipment feasibility, and the regression tree outputs delay prediction. The proposed model demonstrates improved accuracy in predicting delays while simultaneously optimizing resource allocation and rerouting strategies. By automating delay detection and shipment feasibility analysis, the system provides actionable insights that minimize disruptions in supply chain operations. The framework is integrated into a Flask-based web application that enables logistics managers to input shipment details, run predictive analysis, and obtain real-time insights for delay risk, rerouting needs, and resource adjustments, ensuring greater operational efficiency and reliability.