應用重要-表現程度分析與ARIMA 模型在滿意度之分析與預測

The Analysis and Forecast of Service Quality by Using Importance-Performance Analysis and ARIMA Model

劉建璋(liu chien chang)

指導教授 ： 謝俊逸

亞洲大學/資訊學院/資訊科學與應用學系碩士班/碩士(2007年)

摘要

本研究主要分為三大部份，首先第一部份是應用傳統的重要-表現程度分析法(Important-Performance Analysis, 簡稱IPA）及統計分析方法，如變異數分析、相關分析及逐步多元迴歸分析等，試圖找出影響滿意度之因素；第二部份因有些問卷一開始未考慮到重要度資訊，為了解決此問題，本研究將採用滿意度之變異數做為重要度資訊，並且實證以變異數為基之IPA 分析與傳統IPA 分析之間的相異程度；而第三部份主要是採用時間序列分析來延申上一部份的個案研究，以自我迴歸整合移動平均ARIMA 模型(Autoregressive Integrated Moving Average Model) 預測出未來之滿意度及重要度，再將預測結果從事以變異數為基之IPA 分析。
在第一部份的實證是以大專院校學生宿舍做為分析個案；研究結果發現，服務品質構面之「有形性」對於整體滿意度影響程度最大；在傳統IPA 分析中，發現有6 項服務品質是「關鍵優勢」應繼續保持，7 項是屬於「風險與機會」急需改善。第二部份為了比較傳統及以變異數為基之IPA 分析，本研究為了增加可信度，故此部份以大專院校宿舍、校園週邊便利商店及大專院校圖書館三個案例做為實證個案；結果顯示，在大專院校宿舍案例中的28 題服務品質量表有20 題相同，校園週邊便利商店案例中17 題中有10 題相同，最後在大專院校圖書館的案例中19 題有11 題相同，顯示以變異數為基之IPA 分析與傳統IPA 分析之比較，具有一定的相同效果。第三部份則因時間關係，本研究僅選擇以大專院校週邊便利商店之顧客做為時間序列分析的研究案例；分析結果顯示：有2 項服務品質問項落於
關鍵優勢象限，落於風險與機會象限的則有5 項，無任何問項落於潛在威脅象限，落於過度表現之象限的有10 項。此部份研究主要目的為示範應用ARIMA 模型及IPA 分析之方法於服務品質量測的研究是可行的。

關鍵字

服務品質 ； 重要－表現程度分析法 ； 自我迴歸整合移動平均、時間序列

並列摘要

There are three constructs in our study. The first construct is to evaluate the service quality by the traditional importance-performance analysis and statistical methods such as ANOVA, correlation analysis, and regression analysis. The second construct is to estimate the level of importance by weighting methods such as variance methods and entropy method and then to compare the differences among importance-performance analyses when different weighting methods and traditional method are applied to evaluate service quality. The third construct is to predict the level of service quality by the ARIMA(Autoregressive Integrated Moving Average) model. The first construct is engaged in analyzing the service quality of the students' dormitory by using statistical analysis and important-performance analysis. This study intends to find out the deficiencies existed in the students’ dormitory and provide useful information to improve the living and study environment for students. By using the statistical analysis, the results of our study show that 21 out of 28 items in service quality are above average and the most important construct that influences the entire satisfaction is tangible. From important-performance analysis, 6 service items are the keys to gain advantage, whereas 7 items are needed to be improved immediately.
The second construct is engaged in analyzing the service quality of the students’ dormitory in a university, the convenient store around the campus, and the library in a university by variation-based important-performance analysis. This study intends to examine the items of the services by the variation-based importance-performance analysis (IPA) when the importance is unavailable directly from the survey. By IPA, deficiencies can be identified in three different service types and the useful information provided by IPA can be used to improve the service quality for students. The results of our study show that 20 out of 28 items of the students’ dormitory, 10 out of 17 items of the convenient store, and 11 out of 19 items of the library, matched the result of the importance-performance analysis when the importance is available directly from the survey. The third construct has surveyed the same 7-11 convenient store around the campus ten times based on monthly surveys with the same questionnaire to track how the service quality would be changed from time to time. The importance of each item is calculated by its variance, while the performance of each item is calculated by its average. The data were analyzed by ARIMA model to forecast the importance and performance for each item in the future time periods. In addition, by IPA, the existed deficiencies can be identified in a timely basis. The useful information can be provided to improve the service quality for customers. The results show that 2 out of 17 service items are the key to gain competitiveness, whereas 5 service items are needed to be improved immediately. In addition, 10 service items are possible overskill with too many resources on these items. The part of the study demonstrates that applying ARIMA model and IPA in service quality is practical and feasible when time-dependent data are available.

並列關鍵字

Service Quality ； Importance-Performance Analysis ； Autoregressive Integrated Moving Average Model, Time Series

參考文獻

[6] 徐守德、李鎮旗，「企業銷售預測之方法與實証研究 ─ 以台電公司為例表單的底部」，管理評論，13 卷，1 期，23~56 頁，1994。

[21] 應立志、陳福來，「GSP 認證店服務品質之探討－以便利商店業為例」，朝陽商管評論，3 卷，2 期，101~118 頁，2004。

[2] 邱以倫，「顧客對於線上服務品質衡量之研究—以網路銀行為例，台灣大學資訊管理研究所，碩士論文， 2004。

[8] 陳淑女，「陽明山國家公園遊客量之預測─以時間數列模式（ARIMA）分析法表單的底部」，育達研究叢刊，2 期，103~115 頁，2001。

[17] 鄭玫君，「利用以條件機率為基礎之模型於大學校園便利商店之顧客滿意度與銷售預測研究」，亞洲大學資訊科學與應用學系，碩士論文，2006。