台灣鮮乳月產量之分析

Abstract

台灣乳業在一九八Ｏ年後由早期的副業日漸茁壯，更能轉型為中小企業；我們有興趣對此產業的動向及展望深入分析，以每年各個月鮮乳產量為分析資料，透過統計預測以及相關方法，回顧台灣乳業近三十多年來成功發展的歷史。本報告使用時間序列迴歸法、分解法、指數平滑法、ARIMA四種方法估計配適模型，並保留最後12筆資料來做樣本外預測，接著再利用MSE、MAE、MPE、MAPE四個準則來判斷以上四種方法何者較佳，並選出最佳模型；分析工具以SAS統計軟體的預測分析功能為主。研究的最後發現依據預測準則，此資料最適合以指數平滑法-Winters加法模式，本預測結果能提供台灣乳業的經濟策略參考，並提升競爭力。

Taiwan dairy industry started in 1980. We are interested in this industry and would like to model monthly milk productions and to forecast it. We consider monthly of fresh milk production from January 1999 to December 2009. We reserve the last 12 observations for out-of-sample forecasts. We employ four kinds of statistical methods/models for forecasting, which included the time series regression, exponential smoothing, decomposition method, and the ARIMA model. We use MSE, MAE, MPE, and MAPE four criteria to evaluate the above-mentioned four methods to choose the best model. We employ SAS statistical package to analyze data set. The results show that exponential smoothing method – Winters additive model is the best model based on out-of-sample forecast.