Logi Mind Framework: Multi-Dimensional Decision Analytics for Proactive Delay Detection and Operational Optimization

Abstract

Global logistics networks manage billions of shipments annually, yet over 20% of freight experiences delays, and inefficient routing increases operational costs by nearly 15%. Traditional supply chain systems depend on manual intervention or static decision-making, limiting real-time responsiveness and resource efficiency. This study proposes an advanced predictive analytics framework for intelligent delay detection and operational optimization in logistics. The dataset includes detailed shipment records such as delivery timelines, routing paths, resource allocation, and feasibility indicators. A comprehensive preprocessing pipeline comprising data cleaning, normalization, and transformation is implemented to ensure data quality for multi-output predictive modeling. For benchmarking, baseline models including K-Nearest Neighbor with Classification and Regression Tree (KNN-3CA1RT) and Huber-3CA1RT are evaluated. The proposed model, a hybrid Decision Tree-based 3CA1RT (DT3CA1RT), integrates three classification components and one regression module. The first classifier detects rerouted shipments, the second assesses resource allocation feasibility, and the third validates overall shipment feasibility. The regression component predicts delivery delays. Experimental results show that DT-3CA1RT outperforms baseline models in delay prediction accuracy while effectively optimizing routing and resource allocation. By automating delay detection and feasibility assessment, the framework provides actionable insights to minimize disruptions and improve supply chain performance. Additionally, the system is deployed as a Flask-based web application, enabling logistics managers to input shipment data, perform real-time predictive analysis, and make informed decisions on delay risks, rerouting, and resource adjustments, thereby enhancing operational efficiency and reliability.