偏最小二乘法(PLS)简介(Partial least squares (PLS)).pdf

偏最小二乘法(PLS)简介（Partial least squares (PLS) ） Introduction to least squares (PLS) | research report | software | training | knowledge sharing | customer directory | BBS Introduction to least squares (PLS) Number of reading: 14122 release date: 2004-12-30 Jane interface The least squares method is a new multivariate statistical analysis method, which was first proposed in 1983 by s.w. old and c.a. lbano. In recent decades, it has developed rapidly in theory, method and application. Partial least squares For a long time, the boundaries between model and cognitive methods are well understood. And partial least squares rule them organic combine, under an algorithm, can realize regression modeling (multiple linear regression) at the same time, simplify data structure (principal component analysis (pca) and correlation analysis between the two groups o variables (canonical correlation analysis). This is a leap in the analysis of multivariate statistics. The importance of partial least squares in statistical applications is reflected in the following aspects: Partial least square method is a multidependent variable regression modeling method. The least squares method can solve many problems that can not be solved by ordinary multiple regression. The least square method is called the second generation regression method, and because it can realize the comprehensive application of various data analysis methods. The main purpose of principal component regression is t extract the relevant information hidden in matrix X and then to predict the value of variable Y. This ensures that we only use those independent variables and that the noise will be eliminated to improve the quality of the prediction model. Principal component regression, however, still have some shortcomings, as a useful variable relevance is very small, when we select principal component is easy t miss them, making the final forecast model reliability, if we ch se to e