PLS算法的顾客满意度指数模型

通过对国内外顾客满意度指数模型的对比及对我国的现状分析,提出顾客满意度测评的常规模型。在此基础上对常规模型的模型设定、模型估计和模型评价进行研究;利用PLS算法估计模型中的各个参数,将外部估计分为无缺失值和有缺失值两种情况,针对后者提出局部权重新定的缺失值处理方法;以某食品公司为研究对象,对构建的顾客满意度测评常规模型进行实证分析和评价。结果表明:常规模型的显变量符合单一纬度条件、具有一定的解释能力和良好的内敛效度,拟合效果可以接受,测评结果是满意和有效的,为PLS路径模型研究及缺失值处理提供了实用方法。

Research of Customer Satisfaction Index Model Based on PLS Algorithm

Customer Satisfaction measurement regular model is put forward by comparing with Customer Satisfaction measurement model at home and aboard. This paper presents the setup,estimation,assessment of Customer Satisfaction regular model and adapts PLS algorithm to estimate the parameters. The missing data processing method of Partial Weight Setup Newly is proposed to estimate missing values when outer estimate is processed. At last,the system is applied to evaluate some food company and get the standardized path coefficients. The results show that the model meets the Uni-dimensionality,has good expository ability as well as convergent validity and Good-Fit test. The case results prove the feasibility and validity of the model. It offers a useful approach for research of PLS path modeling and missing data processing.