人口政策是政府的重要政策之一，而總人口數則是政府制定政治、經濟、社會及文化發展計畫之主要參考依據，因此如何準確地預測未來的總人口數就成為政府相關部門重要的課題。

本論文試圖為台灣地區總人口數建立時間數列預測模式。我們考慮下列模式：單變量自我迴歸整合移動平均介入模式、時間數列迴歸模式、轉換函數介入模式與指數平滑法，其中轉換函數介入模式中所考慮的投入變數包括育齡婦女總生育率、粗出生率及粗死亡率。我們同時以平均絕對百分誤差 (MAPE) 、根均方百分誤差 (RMSPE) 來評估各模式的預測能力，結果顯示以育齡婦女總生育率為投入變數的轉換函數介入模式最佳，而以粗出生率為投入變數的轉換函數介入模式次之，若以這兩個模式進行未來十年總人口數之預測，並與行政院經建會人力規劃處所作的人口預測中推計值比較，其平均絕對百分誤差分別為0.138%，0.156%，顯示時間數列預測模式有相當佳的預測能力。

In this thesis, we plan to construct various time series models for the total population in Taiwan. The following time series models are considered: ARIMA intervention model, time series regression model, transfers founction intervention model and exponential smoothing method. The input variable considered in the transfer function intervention model include total fertility rate, crude birth rate and crude death rate. We also compare the prediction performance of these models by using mean absolute percentage error (MAPE) and root mean square percentage error (RNSPE). It turns out that the transfer function intervention model with total fertility rate as input is the best model. While the transfer function intervention model with crude birth rate as input ranks the second best. Finally we forecast the total population of the next ten years by using the above two best models and compare with the middle population projection by Manpower Planning Department in Executive YUAN-Council for Economic Planning and Development. The mean absolute percentage error are 0.138% and 0.165% respectively. This result justifies that the time series model has excellent predictive ability and should be considered for total population prediction.

『參考文獻』
[1] 內政部統計處，1998，生命表編算方法改進報告，行政院八十七年研考會經費補助案67-69頁。
[2] 吳柏林與廖敏治，1992，台灣地區結婚率、出生率、人口成長率的時間數列模式探討，人口學刊，第十四期，109-132頁。
[3] 林茂文，1992，時間數列分析與預測，台北：華泰書局。
[4] 鄭天澤與李旭煌，1995，台灣地區出國觀光旅客需求預測模式之比較分析，國立政治大學學報，第七十一期，179-210頁。
[5] 行政院經濟建設委員會人力規劃處，1999，中華民國台灣地區民國87年至140年人口推計。
[6] Ahlburg, D. A. (1992), “Population forecasting：Guest Editors’introduction”, International Journal of Forecasting, 8, 289-299.
[7] Akaike, H. (1973), “Information Theory and an Extension of the Maximum Likelihood Principle”, in Proceedings of the 2nd
International Symposium on Information Theory, 267-281.
[8] Bowerman, B. L., and R. T. O’Connell, (1993), Forecasting and Time Series：An Applied Approach, 3rd ed. Boston：Duxbury.
[9] Box, G. E. P. and G. M. Jenkins, (1994), Time Series Analysis, Forecasting and Control, 3rd ed. San Francisco：Holden-Day.
[10] Carter, L. R. (1996), “Forrecasting U.S. Mortality：A Comparison of Box-Jenkins ARIMA and Structural Time Series Models”, The Sociological Quarterly, 37, 127-144.
[11] El-Attar, S. (1988), “Population Forecasting：An Application of the Box-Jenkins Technique”, American Statistical Association, Proceedings of the Social Statistics Section, 305-310.
[12] Kashyap, R. L. and A. R. Rao, (1976), Dynamic Stochastic Models from Empirical Data, Academic Press, New York, San Francisco, London.
[13] Lee, R. D. (1992), “Stochastic Demographic Forecasting”,
International Journal of Forecasting, 8, 315-327.
[14] Lee, R. D. (1993), “Modeling and Forecasting the Time Series of US Fertility：Age Distribution, Range, and Ultimate Level”,International Journal of Forecasting, 9, 187-202.
[15] Lee, R. D., and Tuljapurkar, S. D. (1994), “Stochastic
Population Forecasts for the United States：Beyond High, Medium, and Low”, Journal of American Statistical Association, 89, 1175-1189.
[16] Pflaumer, P. (1992), “Forecasting US Population Totals with the Box-Jenkins Approach”, International Journal of Forecasting, 8, 329-338.
[17] Schwartz, G. (1978), “Estimating the Dimension of Model”, The Annals of Statistics, 6, 461-464.
[18] Voss, P. R. and C. D. Palit, (1981), “Forecasting State Population Using ARIMA Time Series Techniques”, Technical Series 70-6, University of Wisconsin-Madison, WI.