Theresearch One-commerce Websites Evaluation Based On Statistical Learning Techology

Statistical learning is a branch of machine learning theory based on statistical methods. Now, statistical learning has been successfully and widely used in a varietyof computer applications such as artificial intelligence, pattern recognition, data mining and pattern classification.In recent years, the share of e-commerce websites turnover in GDP has been on the rise, the development of e-commerce has been paid more national attention. The evaluation research on e-commerce website will not only have vital guiding significance to the e-commerce website operators but also have profound significance to the development of e-commerce. Therefore, this paper uses statistical learning methods to evaluate the e-commerce Groupon websites and e-commerce enterprise websites.For the data of e-commerce Groupon websites, we used DEA method and questionnaire survey to calculaterelative efficiency of quantifiable data.Unquantifiable data were fuzzily processed by the method of the fuzzy comprehensive evaluation. And we used PCA method to reduce the dimensionality of the high-dimensional data. Then we used GA for support vector regression parameter optimization, finally we used the optimized SVR for classification and evaluation.For the 100 e-commerce enterprise websites, we used OPTICS-Plus algorithm to cluster the dimension reduction data which was reduced by the PCA method. And we used LLE algorithmto reduce dimension, and a new algorithm called CFSFDP(clustering by fast search and find of density peaks)was proposed to cluster, so as to get the result of e-commerce websites evaluation.Through a variety of methods of classification and evaluation, we analyzed and researched on the features and rules of all kinds of website, and putforward some corresponding proposals for their respective advantages and disadvantages.