運用資料探勘於台灣老年人之分群與分析其和生活滿意度、居住方式的關係

Using Data Mining Techniques for the Clustering of Elderly People in Taiwan And Analyzing Their Relationships with Life Satisfaction and Life Model Factors

林建成(JIAN-CHENG LIN)

指導教授 ： 蔡長明

共同指導教授 ： 柯志鴻(Chih-Horng Ke)

長榮大學/資訊暨工程學院/資訊管理學系(所)/碩士(2012年)

https://doi.org/10.6833/CJCU.2012.00098

摘要

台灣近年來少子化的現象日趨嚴重，且由於醫療科技發達、生活水準提高，人口平均壽命逐漸延長，行政院經濟建設委員會(2010)推估2060年老年人口占總人口比率將高達41.6%，屆時老年人口將成為台灣人口結構中的最大族群。在此種趨勢之下，未來老年人的照料已無法單靠家庭力量來支撐，老年人口的居住問題逐漸浮出檯面，成為政府與社會極需重視課題之ㄧ，有關老年人口居住需求的相關研究實在刻不容緩。此外，老年人口對於居住的生活品質越趨重視，生活滿意度成為其生活品質重要的代表指標。本研究欲藉由老年人選擇居住方式的區隔變數進行老年人口分群，並利用群集分析所得的結果探討不同群集在人口變數、居住方式及生活滿意度上的差異。
本研究為橫斷面研究，以2007年國民健康局「中老年身心社會生活狀況長期追蹤調查」之大規模調查資料作為研究樣本。並結合自我組織圖(Self Organization Map, SOM)、K-means及Dbscan三種群集分析法進行分群，以提高分群的精準度；分群的結果將進一步透過變異數分析、卡方檢定，以研究不同群集資料特性以及各資料屬性間的關係，區分出不同型態的老年人口，並確認不同型態的老年人口之特質、居住方式與生活滿意度之差異。希冀本研究結果可做為政府政策制定與安養中心經營管理之參考。

關鍵字

群集分析 ； 居住方式 ； 生活滿意度

並列摘要

The low birth rate in Taiwan has become increasingly critical in recent years. Due to the improvement in living standard and the advances in medical technology, the lifespan of general population has significantly lengthened. According to the report of the Council for Economic Planning and Development (CEPD) in 2010, the elderly population ratio in 2060 will expect to reach a 41.6% level. At that time, the elderly population will become the single largest age group. As a result, the care of the elderly population in the future will not be affordable for many families. The housing problem of elderly population is expected to become an important issue that the government and the society as a whole must deal with. Today, the elderly population treats the quality of living seriously. Life satisfaction is an important indicator for the quality of living. Hence, the research in housing need for the elderly population is imminent. This study will employ different living styles as segmentation factor to cluster the elderly population. Through the results of cluster analysis, we will discuss their difference with respect to living style and life satisfaction.
This is a cross-sectional study that is based upon the data set of 2007 Survey of the Elderly in Taiwan, Bureau of Health Promotion, Department of Health. This study will apply three well-known clustering methods to enhance the suitability of cluster results. They are self-organizing map (SOM), K-means and Dbscan. We will discuss the meaning of associated descriptive statistics further, and conduct some ANOVA and chi-square test procedures with them to investigate the attributes of the elderly population, and to identify the relationship among them. We hope we will be able to distinguish different types of elderly population, to confirm the characteristics of each elderly population, and to differentiate the degree of life satisfaction among clusters. At last, we hope that the results from this study can offer some helpful suggestions to policy makers in government and to nursing home operators as well.

並列關鍵字

Data mining ； living style ； life satisfaction

參考文獻

李宗派(2004)，「老化理論與老人保健」，身心障礙研究，第2卷第2期，頁77-94。

陳玉敏(2000 )，「長期照護機構老人之適應問題」，護理雜誌，第47卷第6期，頁71-75。

陳榆臻 (2009)，「運用資料探勘中分類技術於預測住院老人跌倒之研究」，元智大學資訊管理學研究所碩士論文。

劉雅文、莊秀美（2006），「探討失能老人家庭選擇長期照護福利服務之決策過程—老人自主權之分析研究」，東吳社會工作學報，第14卷，頁91-123。

中華民國長期照護協會(2012)，「長期照護歷史軌跡」，網址：http://www.ltcpa.org.tw/about/aboutus-304.php。