Enhancing Supply Chain Resilience for Zambian SMEs Using Weather-Aware Machine Learning Application: A SEM Analysis of Forecasting Accuracy and Business Outcomes Influences

Abstract

This study investigates the integration of weather-based machine learning (ML) models in improving supply chain resilience and market forecasting accuracy for small and medium enterprises (SMEs) in Zambia. The purpose is to examine how weather-integrated forecasting systems enhance demand prediction, inventory optimization, customer retention, and marketing strategy refinement in the SME context. Using survey data from 293 SME stakeholders, the study employed structured equation modeling (SEM) to analyze the relationships among weather patterns, forecasting accuracy, system usability, and business outcomes. Findings indicate that SMEs leveraging weather-enhanced ML models achieved a 63.8% improvement in order prediction accuracy, enabling more responsive and data-driven decisions, with some recording a 46.1% increase in order predictability accuracy and 61.4% improvement in spotting market opportunities. The total effects on market opportunity identification and system usability for business planning showed a strong positive relationship. The study also explored correlations between weather patterns and market characteristics, linking item popularity to system forecasts. An intuitive application for SMEs was developed to capture and visualize market forecast data, focusing on user comfort and usability challenges. By embedding analytics into operational workflows, the research contributes to a practical and replicable framework for adaptive supply chain management and resilience building in emerging markets. This study is among the first to integrate weather-based ML forecasting with SEM analysis for African SMEs, offering original insights for data-driven decision-making and sustainable business growth.