Research On A Number Of Issues Related To Partial Least Squares Regression Analysis

In the multivariatc regression models, multicollinearity between the multivariate are very common. For the problem of multicollinearity and the number of explanatory variables rather than practical issues, S.Wold and C.Albano first proposed Partial Least Squares Regression(PLSR) in 1983. In this paper, we discussed some issues of PLSR. First,Partial Least Squares Regres-sion are given theoretical derivation. Second, we compared the relationship between Partial Least Squares algorithm(PLS) and Principal Component Analysis algorithm (PCA). Obtained: Partial Least Squares iterative algorithm to handle data sheet leaflet extraction equivalence with Principal Component Analysis. Finally, we discussed the issue of fixed-order about PLSR. According to a given degree of freedom of several Partial Least Squares Regression unbiased estimates,and through data simulation, we compared several regression models which were selected based on information criteria.