有關時間序列資料的分析方式，為了簡化估算程序，通常將變化趨勢假設為線性增減模式。本研究嘗試將物理學門發展的混沌理論，跨域應用於檢驗旅遊人數時間序列是否具有混沌（非線性）變化。基此，選定南投縣清境農場、惠蓀林場、鳳凰谷鳥園及溪頭自然教育園區為研究對象，並且應用Rosentein, Collins 與De Luca（1993）提出之「小數據集」方法，分
析該農場由1998 年1 月至2010 年9 月（合計153 個月）之旅遊人數時間序列資料。研究發現清境農場及溪頭自然教育園區的旅遊人數時間序列變化應存在混沌特性。故對於類似旅遊
人數時間序列資料仍應進一步考慮以非線性時間序列方法分析，方能得到較為客觀與準確之成果。 For estimation simplicity it is often assumed the trend as linear type for time series analysis. This study aims to apply chaos theory of physics science to the topic of visitors forecast in social science and tries to examine chaos (nonlinearity) in these time series data. Cingjing Veterans Farm, Hueisun Experimental Forest, National Fonghuanggu Bird Park and Xitou Nature Education Area in Nantou County is selected as a case study and “small data set” method proposed by Rosentein et al. (1933)
is used to analyze visitors’ time series data from January 1998 to September 2010 (153 months).
The results show time series variation of Cingjing Veterans Farm and Xitou Nature Education Area maybe exist chaos (nonlinearity). Hence these chaos type time series data are required to consider nonlinear method or model to improve the accuracy of forecast.
