Modeling and forecasting the fabric demand in apparel industry: a statistical approach

Abstract

The competitiveness in the apparel industry is continuously accelerating, thus, demand forecasting plays a vital role in decision making. The effectiveness of the decisions helps to increase profitability and customer satisfaction by reducing the risks in business activities. A well-recognized garment manufacturing company which supplies garments for a world-renowned sports brand was selected in this study. On Time Performance of the factory was found to be considerably reduced by over-forecast or below-forecast and hence influence the rating of the Factory’s performance. The main aim of this study was to model and forecast accurate future demands for selected fabric category. Secondary data was collected for the study using bi-weekly actualized customer demand and the sample covered data from December 2019 up to February 2023, summing up to 76 data points. Minitab software was used for descriptive data analytics and R statistical software package was used for the time series modelling. According to the descriptive data analysis, year by year, the demand was increasing exponentially. The summer season demand was significantly lower than that in other seasons. When it comes to the monthly demand, April, January and August months showed the highest demand and considerably low demand in February, October and November months. If buy-wise demands were considered, April 2nd, August 2nd, June 2nd, January 2nd and April 1st buys have the highest demands respectively. Demand is very low in February 2nd and October 2nd buys. Among the fitted ARIMA models the ARIMA(4,1,4) model with non-zero mean was identified as the best-fitted model. Model diagnostics confirmed that the selected model is well-fitted. The results indicate that the forecasting performance of the selected model is highly accurate with expected accuracy level for the immediate next 3 buys with a mean average percentage error of 6.85%.