基于数据色彩与文法Fuzzy识别的回归模式选择

依据人们对数据的几何形态来推断变量间函数关系的原理，提出了色彩数据与Fuzzy嵌入文法等概念，并以此建立了利用计算机自动选择回归方程的文法Fuzzy模式识别方法，整个模式识别的过程完全是一种模拟人的经验识别过程，识别效果的好坏，主要取决于两个主要手段，一是色彩滤波，一是模式的文法规则，色彩重心是将原始数据上赋予不同权重(色彩度)条件下得到的，信息损失的少，更接近于人为描述曲线的心理特征，回归方程模型由一组描述曲线增减变化趋势的语言变量组成。

Choice of regression mode based on data color andfuzzy grammar identification

According to the principle of deducting the function relationship among variates from the geometric figure of data,the concept of color data and fuzzy imbedding grammar etc is put forward,whereby a method of identifying the fuzzy grammar,by which computer can choose regression equations,is established.The whole process of identification is a kind of process that imitates the experience identification of human.The quality of the identification result is based on two main conditions:the one is color filtering and the other is grammar's rule.We get the color gravity center by giving the original data different weight number.The less information loses,the more near to the psychological character of people's describing curves.The model of regression equations is composed of a set of mode linguistic variates,by which we deseribe the curver`s direction of increase of increasing or decreasing.