Demand Forecasting System for Men’s Apparel Sector in Sri Lanka

Abstract

This research is conducted with the aim of developing a sophisticated demand forecasting system for the men's apparel industry in Sri Lanka using advanced machine learning techniques. The varieties estimated in this research are ARIMA, ETS, Linear Regression, Random Forest, Gradient Boosting Machine, and Long Short-Term Memory Networks against a dataset enriched with sales data and economic indicators such as GDP, inflation, and price. Our results indicate that the GBM model has the best performance in terms of attaining the smallest error metrics, which proves its efficiency in capturing complex and nonlinear relationships in data. Model accuracy is improved when additional economic indicators are added for evaluation, thus proving their critical role in the demand forecast. These research findings are very resourceful to the industry players to help them better plan their production, inventory, and react to changeable markets. However, hybrid model development, integration into real-time data, and further exploration of advanced neural network architectures are identified areas of future work. In the present study, besides advancing the literature related to machine learning applications in the apparel industry, solutions have been provided to seek operational efficiency and profitability.