Machine learning-assisted efficient demand forecasting using endogenous and exogenous indicators for the textile industry

Abstract

Demand forecasting is quite volatile and sensitive to several factors. These include firm-specific, i.e., endogenous as well as exogenous parameters. Endogenous factors are firm-specific, whereas exogenous factors are the macroeconomic indicators that significantly influence the demand forecasting of the firms involved in international trade. This research study investigates the significance of endogenous and exogenous indicators of demand forecasting. For this purpose, we use daily production data from a textile apparel firm for the period from May 2021 to January 2022. In the first step, we employ generalized least square and single-layer perceptron models for coefficient estimation to investigate the impact of each indicator. In the second step, we use linear regression (LR), support vector regression (SVR), and a long short-term memory (LSTM) model for demand forecasting. The forecasted results using SVR and LSTM reveal that errors are reduced when exogenous indicators (exchange and interest rates) are used as inputs.