Demand Forecasting Using Artificial Neural Network Based on Quantitative and Qualitative Data

Abstract

Demand forecasting plays a very significant role in logistics activities due to efficiently guiding various activities and increasing the commercial competitive advantage in a steadily fluctuating business environment. Traditional methods generally employ statistical methods, such as exponential smoothing or auto-regressive integrated moving average methods, to forecast the product demands. However, most researches only consider quantitative data such as time series data. Despite the fact in the real world, there are diverse qualitative data have more effect on product demands. Hence, this study aims to propose different 3 models based on quantitative and qualitative data. Such models consist of model 1, used only qualitative data as input, model 2, used only quantitative data, and model 3, used both of qualitative and quantitative data. A TV demands data set provided by a well-known shopping mall in Thailand is adopted to verify the proposed model. According to the consideration of performance measurement, the finding indicates that model 3 is outstanding. It can be concluded that the qualitative data are very important to demand forecast.