半導體IC產品的市場需求較不穩定,若每月實際庫存量與預測數字相差過大,其存貨成本容易上升,如果可以做出準確預測,可避免不必要的損失,雖然預測不能百分之百正確,但可替經營管理者提供一具有科學化的庫存預測。本研究擬使用AweSim 模擬工具以科學化的預測模式來預測半導體IC設計業庫存量,探討影響IC設計業庫存量主要變數,以及利用以往庫存資料,建立適當的模型來預測未來庫存量。 在此研究中,透過實例的模擬驗證,加上模擬預測的分析方法,可以顯示出應用Demand pull的觀念及方法並進行適度調整目標庫存水準。模擬系統程式所建構的預測模式是可依據使用者不同的需求(如修正週數)及模式修正方法(不修正、指數平滑、加權平均、指數趨勢等)持續進行修正。分析結果發現,所有模式預測評估指標的MAPE平均值均小於0.3%,可見模擬系統程式本身的預測效果已經是相當理想。其中以指數趨勢修正部份在帳面庫存值與實際庫存量部份是較為接近的,故其預測結果是比較理想的。
The uncertain market demand of semiconductor IC products makes a company rise its inventory cost if the demand forecast cannot work properly. If a company can make accurate forecasts, it might avoid some parts of unnecessary losses. Although the forecast is not one hundred percent correct, a manager still can catch sufficient information to manage the product inventory. This study intends to build a forecasting model by using AweSim simulation tool to predict the semiconductor IC design industry inventory levels, to explore the impact of the main variables for the IC design industry inventory, as well as to operate sales process with the predict data. In this study, through simulation and verification of the analysis and forecast, it can be shown the model with demand pull appropriately adjust the target inventory levels. The predict model constructed by the simulation system program can be tuned by different user needs (such as the correct number of weeks) and modes of correction methods (but not to amend the exponential smoothing, weighted average, exponential trend) continued to be corrected. The analysis results showed that all models predict the evaluation index MAPE average of less than 0.3%, visible simulation system, the program itself prediction is quite satisfactory. The forecasted inventory value by the exponential trend is closer to the actual value, so the forecast is more ideal.