摘要
用户满意指数能反映出影响被测评对象满意程度的各个因素之间的本质关系,从而为企业经营管理提供可靠的信息。根据建筑施工企业自身特点,提出了建筑施工企业用户满意指数模型,构建了各个隐变量所对应的观测变量,给出了模型结构分析的数学表达式。针对模型隐变量之间存在的多重相关性和样本点个数与变量个数相比明显过少的问题,采用基于主成分和典型相关分析的偏最小二乘法(PLS)使指数模型的求解得到了很好的解决,用PLS路径分析法来估计模型隐变量和显变量对相应隐变量的解释程度系数,用PLS回归分析法来估计模型隐变量之间的影响系数。通过对成都市某建筑施工企业进行用户满意指数测评,对测评结果的隐变量之间进行了详细的相关关系分析和回归分析,为企业实施用户满意战略、提高用户对建筑施工企业的满意程度提供了积极有用的信息。
CSI(short for Customer Satisfaction Index) can reflect the essential relationship between each factor which has impact on satisfaction degree of evaluated object,thereby,to supply reliable information for management of enterprises.According to the specific characters of construction enterprises,by taking the evaluating model of the customer satisfaction index at home and aboard,model of customer satisfaction index for construction enterprises is established,observed variables of each latent variable are proposed,and the specific mathematical expression of the model is put forward.In order to solve the problems of multiple correlations among the factors in the model and the obviously inadequate quantity of samples compared with the quantity of variables,this paper presents the way of partial least squares(PLS) based on principal components and typical correlative analysis.The latent variables and the explanation degree coefficient of the manifest variable to the corresponding latent variables are estimated by PLS path analysis,and the influence coefficient among the latent variables in the model is estimated by PLS regression analysis.The evaluating results of the Customer Satisfaction Index measurement carried out in a construction enterprise in Chengdu,Sichuan Province,provide useful information for construction enterprises to carry out customer satisfaction strategy and for the enhancement of the customer satisfaction to these enterprises.
出处
《西南石油大学学报（社会科学版）》
2009年第1期48-58,共11页
Journal of Southwest Petroleum University(Social Sciences Edition)
基金
国家社会科学基金(06XJY003)
关键词
用户满意指数
建筑施工企业
偏最小二乘法
影响系数
customer satisfaction index
construction enterprise
partial least square
influence coefficient
