Demand Forecasting for Perishable Products Using Data Mining Techniques: Systematic Review

Demand forecasting for perishable goods has become increasingly critical due to the growing complexity of supply chains and the need to minimize waste. This systematic review focuses on cold chain logistics products, excluding agricultural items, to identify key challenges and significant demand factors and to evaluate the effectiveness of traditional and advanced forecasting methods. To achieve these objectives, the PRISMA framework guided the selection and analysis of 54 studies based on specific inclusion and exclusion criteria. Key findings highlight consumer preferences, income, and product pricing as major factors affecting forecasting accuracy. Traditional models like ARIMA and Holt-Winters are widely used for their effectiveness in capturing trends and seasonality but struggle with outliers and sudden demand shifts. Advanced techniques, such as LSTM, are effective at identifying detailed and complicated patterns but demand significant computing resources and large datasets. The findings underscore the need for hybrid approaches combining traditional and advanced methods to improve forecasting, enhance supply chain efficiency, and reduce waste. This research offers actionable insights for optimizing cold chain logistics and strengthening the perishable goods industry’s competitiveness.