台灣地區水果銷售預測研究

A Study of Fruit Sales Forecast in the Taiwan Market

吳春慶(Chun-Ching Wu)

國立虎尾科技大學/管理學院/工業工程與管理研究所/碩士(2006年)

摘要

台灣於二OO二年加入WTO，打開國外農產品進口及降低關稅大門。加入了WTO雖使國內農產外銷機會大增，但也因開放農產品之進口，造成國內農產市場價格低迷不振、供需失調，使農業淪為弱勢產業。依據行政院農委會調查報告，水果為台灣地區極具高經濟價值之農產品(水果產值約佔全國農業總產值之百分三十九)[63]，因此不少農民紛紛轉向水果的種植，但也因此使得水果種植面積的不斷擴張，造成產銷失調 [61,63]。
故生產者應依據市場需求，適時調整水果產量，力求供需平衡以達最高生產效益，因此本研究欲針對台灣水果市場進行銷售量之預測，期能找出最適預測模型以作為果農生產依據。在此本研究所挑選之水果別係以市場最具銷售潛力之西瓜、香蕉、鳳梨及梨作為探討對象(四項水果約佔全國水果總產值的27%)[63]。
本研究所採用的預測方法為定量分析法中的古典分解法及人工智慧法中的灰色預測模型，並將季節因子導入傳統灰色模型而成灰色季節模型。在此，本研究亦針對各預測模型提出殘差修正，其中包括古典殘差修正模型、灰色殘差修正模型及灰色季節殘差修正模型。
本研究採用91年第一季至94年第二季各項水果歷史銷售資料作為模型輸入之用，經實證分析可得香蕉銷售之最佳預測模型為古典殘差修正模型，其它水果項(西瓜、鳳梨、梨)之最佳預測模型為灰色季節殘差修正模型。本研究發現當銷售值具有趨勢或季節波動時，傳統灰色預測模型無法求出較佳預測精確度(即較不適用)，而具有季節及趨勢因素導入之古典殘差修正模型、灰色季節殘差修正模型可得較佳預測效果。故本研究認為當銷售數據呈現週期性循環時，唯有將季節、趨勢因素導入才能求出較佳預測精確度。
在求出各項水果之最佳預測模型後，本研究並運用馬可夫觀念進行九十五年度各季水果銷售之推估，期能作為果農生產銷售之依據，以降低長鞭效應之影響，使得水果承包商、零售商及果農皆能得到更高的經濟效益。

並列摘要

Sales forecast can help decision-makers not only to predict the market demand, but also to provide the basis of the optimum product policy.
In the thesis, we studied fruit sales forecast in the Taiwan market. Four kinds of major fruits, including watermelon, bananas, pineapples and pears, were discussed. There are six proposed forecasting models: the first model is the seasonal variations model; the second one is the GM(1,1) grey forecasting model; the third one is the grey seasonal variations model; the fourth one is residual-modified seasonal variations model; the fifth one is residual-modified GM(1,1) grey model; and the six one is residual-modified grey seasonal model. The sale volumes reported in the Agricultural Products Market Information System [15] were used as test data sets.
The result shows that the residual-modified seasonal variations model can provide the best forecast in the banana case, and the residual-modified grey seasonal model can provide the best forecast in the other three cases. Finally, the best models were used to forecast the sales of 2006, and the results are hoped to provide the economic values to the related members.

參考文獻

42. Box G.E.P. and Jenkins, G.W., 1970, Time Series Analysis: Forecasting and Control, Holden-day.

43. Brown, R.G., 1963, Smoothing, Forecasting and Prediction of Discrete Time Series, Englewood Cliffs, N.J.: Prentice-Hall, Inc.

44. Chang, S.C., 2005, “TFT-LCD industry in Taiwan: competitive advantages and future developments”, Technology in Society, vol. 27, pp.199-215.

45. Deng, J.L., 1982, “Control problems of grey systems”, Systems and Control Letters, vol. 5, pp.288-294.

46. Ho, S.L., Xie M. and Goh, T.N., 2002, “A comparative study of neural and Box-Jenkins ARIMA modeling in time series prediction”, Computer and Industrial Engineering, vol. 42, pp.371-375.