| Popis: |
Though quality control is implemented at manufacturing side, it is still unavoidable to ship defective products not detected to the customers. When defective products are detected at customer side, the product returning processes will correspondingly activate. However, it leads to the increase of operating costs that manufacturer side should make a certain amount of stock to solve the situation of returning products. Accordingly, the tradeoff is between over-stock and under-stock to balance the cost-cutting and customer satisfaction. The grey models (GMs) are widely applied in short-term time series data prediction; though, the improvement of prediction still exists. Hence, this research reveals a new GM, which employs the mega-trend-diffusion technique, to estimate the background values in the traditional GM. A case is studied from a leading TFT-LCD company in Taiwan. Comparing with two GM models, the proposed model outperformed concerning the case data. |
