The Point Transfer Method Study In Interval Comprehensive Evaluation

The method of comprehensive evaluation has been penetrated deeply into all aspects of the social and economic. As an important branch of economic statistics, comprehensive evaluation gets increasing concern of the community. It has become an effective yet popular tool using into all kinds of assessment, appraisal, identification recognition activities. The main feature of comprehensive evaluation is a combination of quantitative and qualitative analysis. By study the statistics of phenomena, we can have a deeper and more comprehensive understanding of the object. At the angle of statistical activities’ purpose, comprehensive evaluation is a significant step based on statistical survey and data processing. It gives us the comprehensive scale of the object we studied.In the situation of using the method of traditional comprehensive evaluation, the data format is the form of point value. However, due to the characteristics of the evaluation methods, different methods of data structures and evaluation model has different requirements and regulations. Also while using comprehensive evaluation, fuzzy data and data uncertainty is common cases. So the form of interval value appeared. How to carry out evaluation activities when the data is interval value, has become one of the issues we need to consider.Comprehensive evaluation of interval data using, however, there are series of questions. First in holistic ideas Interval Evaluation, because of the characteristics of interval numbers it requires the development of some new comprehensive evaluation techniques which almost abandoned the traditional evaluation techniques. I believe that the idea of "transformed" of interval number is more in line with the actual operation. After "transformed", the data structure can be used with traditional methods of evaluation, which makes the evaluation easier. Therefore, the entire interval number corresponding evaluation becomes an issue of finding a reasonable way to shift interval values into point values. This paper proposes a way to find point value similar as determine the physical center of gravity.The writing scheme of this article is to divide interval number into two types:distribution information known and distribution information unknown, then handle the two cases respectively. Under the distribution information known condition, this paper introduces a physical center of gravity ideological to shift interval information into point value. Under the distribution information unknown condition, this paper proposes a same indicators distribution similarity assumption, with this assumption to estimate the distribution of interval number, then as the distribution information known condition to finish the point value transformation.The chapters’ arrangement is as follows:Chapter one expounded the basic problem of interval evaluation techniques. About the interval numbers generation, the type of interval number and interval number handling ideas, explore the point values of the interval indicators treatment as a comprehensive evaluation of the feasibility, and lay the foundation for the entire piece article.Chapter two describes the interval symbolic data computing, focusing on the statistical description of interval symbolic variables, including the experience of the interval number density function to calculate the mean and variance, covariance and correlation coefficient calculated and comprehensive evaluation system of interval number of indicators to quantify indicators standardized.Chapter three assumed that the variable distribution is arbitrary, i.e. the distribution of the evaluation unit in the same variable indicators may be different on the distribution of the different variables of the same evaluation unit interval can also be different, and distribution can be in the form of biased can also have a peak, including multi-peak situation under the ideal case that distribution information is completely known and Introduced a physical center method to shift interval information into point value.Chapter four assumed that the distribution information is unknown, but the same indicator has a similar focus on the distribution. In this chapter, we stick on single-peak situation. By using the theory of phase space reconstruction to send up individual indicators matrix of interval numbers and get more information. Due the single-peak may have partial peak, so using Beta-distribution to estimate distribution of interval numbers.Chapter five considered there are several peaks within the interval indicators. As the characteristics of several-peaks, this article proposed two possible treatment methods:1. direct estimate the distribution;2. separation peaks then doing single-peak estimate.Chapter six is about the summary and outlook.