| E-Business has been a main route for corporation to grow rapidly and be more competitive. The overall performance of e-business platform reflects companies' services and competitiveness. How to enhance companies' e-business performance through the e-business performance evaluation becomes the emphasis of the government, academe and enterprise's research. And designing, applying effective and practical evaluative algorithm is the key area of the researches concerned. In this paper, based on the previous multi-layer algorithm architecture, the multivariate statistical analysis is imported. First, with the cluster analysis, the evaluation indexes were divided into several classes with the objective relationship among them. Secondly, the principal component analysis pick-up the key indexes, with the new integrate indexes, the analysis dimensionality was reduced. Then, some answer sheets which violate the objective relationship of the indexes were wiped off by using the canonical correlational analysis. After the above steps, the new answer sheets were used as the data source of the AHP method. With less scrap answer sheet, the analysis result was more stable and accurate. Besides, the single variant analysis upgrade to multivariate analysis made the synthetically evaluation of experts more comprehensive and scientific.This paper made a test of the new algorithm architecture to prove that the intelligent analysis platform which based on the new algorithm had improved the evaluate ability. This paper described the mode and implementation of the new algorithm architecture. As a result, combing the entire formal evaluation process, ten most typical samples of comprehensive e-commerce are selected, and therefore related assessments have realized all-rounded. Though the contrast of the evaluate result, more stable and objective of the new multivariate analysis algorithm compare with the previous version was proved and visualized. |
PDF Full Text Request