반도체 소재-부품 공급사의 인공신경망(ANN)을 이용한 수요예측 연구

초록

Global companies need to increase their competitiveness through efficient management and rapid decision-making of limited internal resources. Supply chain management(SCM) expands from raw material procurement to final customers. Due to the limitations of corporate resources and capital, accuracy of demand forecasting is essential to improve competitiveness. This study tested the artificial neural network(ANN) model to use it as a more improved demand prediction model required by material suppliers in the semiconductor industry. The predictive results of artificial neural network models and other traditional time series models such as Moving Average, Exponential Smoothing including Holt-Winter’s model, and ARIMA were compared with actual sales data. Based on this, the accuracy of the artificial neural network model according to the demand pattern of the semiconductor component industry was evaluated. The artificial neural network model predicted the highest average accuracy rate among demand prediction models and it can be expected to improve overall demand forecast accuracy which is able to contribute actual facing problem at similar supply chains. Demand forecasting is a beginning of sales and supply planning in supply chain management. This study is expected to provide one of cases at the demand forecast research due to lack of domestic research papers. It will also provide practical guidelines of real-world problem at various companies which was used actual data from semiconductor industry.

키워드