The fashion products are sensitive to seasons and fashions, and have a short life cycle. Also, their product values drop rapidly as time goes on. As a result, the accurate forecasting of demand is very important. In this study, comparative analysis is...
The fashion products are sensitive to seasons and fashions, and have a short life cycle. Also, their product values drop rapidly as time goes on. As a result, the accurate forecasting of demand is very important. In this study, comparative analysis is performed to the fashion products for selecting the most appropriate forecasting methods. Among the forecasting techniques, only the time series models are considered such as decomposition method, single exponential smoothing, Holt's model, Winters' model, and Box-Jenkins’ ARIMA model. The actual data from a fashion company are gathered and they are from five product groups which are representative to the typical fashion products. Applying the time series forecasting methods, the accuracy is measured by MSE, MAD, and MAPE, respectively. As a result, the Winters' model shows the least error for all five product groups. The Winters' model gives the more weight to the recent data while reflecting a change in level, trend, and seasonality. These properties are proved to be appropriate to the characteristics of fashion products. The result of this study will be able to be utilized in order to improve the demand forecasting in the fashion company.