<![CDATA[The e-commerce logistics industry faces challenges due to the high risk of fleet delivery delays, where the analysis of risk mitigation and its financial impact is often limited for retail operational management. This project aims to develop an Integrated Logistics Risk Management Decision Support System Smart Platform named LogiTrack v2.0, which combines two main data engineering domains: Data Mining and Machine Learning. For data analysis, an artificial intelligence model was developed using the Random Forest Classifier algorithm within a Scikit-Learn Pipeline architecture, which detects minority class data via the class_weight='balanced' parameter to automatically predict cargo delay status based on the engineering of four composite features (Volume, Density, Shipping Fee per Gram, and Route). For smart interaction, binary file .pkl optimization was implemented using the Joblib library with compression level 3 to radically reduce storage capacity from 138.80 MB to 28.40 MB. All of these functionalities are integrated into a single interactive web application built using Streamlit and deployed on the Streamlit Cloud server. The dashboard is equipped with two main tools: a Guardrail Engine function to block package inputs exceeding retail capacity ($>30$ Kg) and a Business Impact Calculator panel to calculate the conversion of management financial penalties by 20% in real-time. Functional testing results via Black-Box Testing show that the classification model achieves a global accuracy rate of 90.00% and a precision value of 0.31, while the web platform successfully operates stably with low latency. This platform offers a holistic solution that enhances cargo distribution transparency, optimizes operational risk control efficiency, and suppresses potential financial losses within the logistics industry ecosystem.]]>
