決策樹應用於連續性照護退休社區之市場區隔基礎研究

社會變遷導致高齡者對於長期照護機構的需求逐年提高，加上臺灣現有之長期照護機構不論社區式或機構式皆有其缺失，故連續性照護退休社區提供了高齡者除社區式與機構式照護外的另一種退休安排的選擇。國內目前針對連續性照護退休社區的市場區隔研究尚未涵蓋完整之服務設施項目，本研究目的以決策樹演算法找出市場區隔及有效區分市場區隔之區隔基礎，以瞭解完整的連續性照護退休社區所應提供的服務類型，並提供有意在臺灣經營連續性照護退休社區之企業，有相關資訊及模式參考。本研究以人口統計變數、連續性照護退休社區提供之五大類服務設施問項為自變數，經資料前置處理後，選取三項不同的依變數，利用決策樹建立預測模型，得出市場區隔及市場區隔變數，並以卡方檢定與Schiffman and Kanuk學者提出之市場區隔準則檢定區隔變數的強健性。實證結果顯示，若以「入住意願」為目標，則最佳市場區隔變數為「影響入住選擇因素區隔」、「財務協助之工作抵費」、「不動產擁有狀況」及「每月可支配所得」。若以「影響入住選擇因素區隔」為目標，則最佳市場區隔變數為「入住意願」、「子女狀況」、「獨立自理能力」、「不動產擁有狀況」及「目前住處之房屋所有權」。
Global population is growing older, and the elder market becomes an important segment gradually, especially the needs for long-term care. In Taiwan, a new model, “Continuing Care Retirement Community (CCRC),” is proposed for the retired people. In the previous research, there was no any variable about the services and facilities within a complete CCRC used for segmenting the CCRC market in Taiwan. Therefore, the goals of our research are not only to understand the types of services and facilities a complete CCRC should provide, but also to use decision tree in finding the valid bases of market segmentation that can effectively segment the CCRC market in Taiwan. We analyze the result and provide suggestions for the enterprises that want to operate the CCRC business in Taiwan. We take the demographic variables and variables about the five types of services and facilities a CCRC should provide as independent variables. There are three cases for decision tree induction, for each case we use different dependent variable. After building the tree-like prediction models, the bases of market segmentation which are took out from the tree model must be testified by Chi-square test and evaluated by Schiffman and Kanuk's criteria. The results are as the follows. If “the will to live in a CCRC” is the dependent variable, the best segmenting variables are “factors effect the will to live in a CCRC (clustered),” “financial support-working,” “owned realty,” and “monthly income.” If “factors effect the will to live in a CCPC ” (clustered) is the dependent variable, the best segmenting variables are “the will to live in a CCRC,” “children,” “self-conducting,” “owned realty,” and “residence ownership.” dimension of service quality is not revealed positively associated with customer loyalty. The corporate image and relationship quality do play the mediator roles between service quality and customer loyalty.