The Evaluation Model Of Complex Data Multi-attribute

Complex data attributes problem is a kind of phenomenon of people at work and daily life common often encountered, so for complex data attributes issues of research is positive significance. For multi-attribute questions research started in 1957, when Ackoff. Churchman. Arnof etc, firstly official using simple weighting method to deal with 'choice enterprise investment policy "multiple attribute decision making problems. But at the time of multiple attribute decision making problems with the research does not cause the attention of people. Until the 1960s, with decision-making science development and practical the needs of production and life of multiple attribute decision making problems research gradually got the attention of people, become decision-making science research hotspot.This article from multiple attribute index, and combining relevant statistical knowledge and related theory for complex data of the stratified sampling processing, simplified data, on the basis of data clustering to find the inner relationship between study object, and with the improved cellular automata, the improved genetic algorithm of the intelligent algorithm is established, and the estimation of model proposed model is applied to estimate China Shanghai a-share stock movements and China's high-tech workers social situation, through empirical findings estimate model improves the convergence, and effective estimated the research object of the trend.The innovation of this paper is as follows:(1) In the process of weight coefficient, this paper puts forward a new method, which integrates the objective method and subjective method, it is effective to prevent objectively and subjectively one-sidedness, to make the result more actually.(2) In the process of the weight coefficients, this paper clustered stocks, get the clustering attract rate and coefficient, and using the improved genetic algorithm, which provided the basis for the objective method and subjective method.(3) According to multiple attribute index, we improved the fitness function, mutation probability, crossover probability of the genetic algorithm, empirical proves that the conclusion is not only more accord with the objective reality of attributes, and effectively avoid the "precocious", make sure that the search process can jump the local optimal and speed the convergence.