Based On DEA's E-commerce Site Quality Diagnosis Model And Algorithms

The quality of e-commerce website directly impacts on competitiveness and economic benefits of enterprises, and thus how to diagnose website effectively and provide continuous improvement method has become an important study topic.This paper reasearches quality diagnosis model, solving algorithm and other issues based on DEA (Data Envelopment Analysis). The main results of the thesis are as follows:Firstly, the quality diagnosis model and system of evaluation target based on DEA are proposed.This paper builds a mathematical model named ECWQDM based on DEA for diagnosing the quality of e-commerce sites.The input-oriented and output-oriented production models and BCC model which allows for variable returns to scale in DEA models provide the basic theory for the proposed model. According to this model, we can diagnose fundamental and local factors impacting the quality of websites from the microscopic point. Moreover, this paper, creating a number of input and output indictors for each decision-making unit based on the user’s access behavior, defining each category of goods in e-commerce website as a decision-making unit which has lots of input and output indicators. We propose the evaluation metrics combing the established quality diagnostic model from the view of diagnosing the effectiveness for each commodity category in e-commerce website as well as the inefficiency of main functional pages. As a result, a workable and high credibility system of evalution target for diagnosing quality of e-commerce website has been established.Secondly, the solving algorithm of above-metioned quality diagnosis model is given based on the idea of saddle point algorithm. In response to the above-quality diagnosis model with strongly degenerate linear programming characteristic, the use of traditional methods to solve the above model is very difficult. This paper designs ECWQDSPA algorithm for solving the above-metioned quality diagnosis model based on the idea of saddle point algorithm. Finally, the ECWQDM model and the ECWQDSPA algorithm are verified to be feasible and effective by the simulation experiments of B2C e-commerce websites.Finally, a system prototype of e-commerce website quality diagnosis is designed and implemented. In this thesis, the system prototype of e-commerce website quality diagnosis is designed and implemented with Matlab+Jsp+MySqL technology, and testing the model and algorithm on basis of an e-commerce website. Through experimental analysis, validates the feasibility of the ECWQDM model and ECWQDSPA algorithm. Users can make use of the system prototype to diagnose the quality of e-commerce website according to the needs of oneself. The results of this thesis can provide a reference for website managers to diagnose the quality of website.