全球半導體市場銷售趨勢預測研究─以r2 增量指標探討

Abstract

本研究是根據全球半導體市場歷年來銷售額數據，利用迴歸分析尋求一個數學模式，來預測全球半導體未來中短期銷售額的發展趨勢。 利用迴歸分析中，簡單線性迴歸模式，及幾個常見的簡單非線性迴歸模式，進行迴歸分析獲得判定係數 r2以及預測值。比對歷年來銷售額數據與預測值的誤差值，選擇誤差值比較小者，做為預測未來的迴歸模式。二次多項式迴歸模式，是最適合做為預測未來，全球半導體年銷售額的迴歸模式。並以Δr2值(r2值增量) ，用來當作迴歸模式，外插預測結果優劣的參考指標

Regression analysis method is used to study on the whole wide semiconductor sales forecasting. Different partial historical sales data generate some different regression models and predict data. After the error compression between the predict data of the regression models and the actual historical sales data. The quadratic equation model is the best regression analysis model for future forecast of the whole wide semiconductor sales. The increased trend of the coefficient of determination r2 has a good relation to as the determinant for the sales forecast.