Linear Regression Models for Assessing the Ranking of Web Sites Based on Average Numbers of Visits

As the World Wide Web develops, the number of new websites grows at an astounding speed. However, many web sites are designed and established without the aid of sufficient professional skills and resources, so their quality is often questionable. Therefore, evaluation of the quality of web sites becomes important. In this work, stepwise regression method is applied to assess the ranking of web sites in two different categories based on the average number of visits per day. Such ranking represents, in part, user satisfaction, and therefore, the quality of a web site. A total of fourteen factors frequently found in the literature are considered independent variables in developing the model. The regression analysis shows the regression models are different for two categories of web sites and their prediction accuracies are both over 75%. Our models also suggest that good web sites must satisfy a wide spectrum of quality aspects.