An Interpretation of Partial Least Squares

摘要: Abstract Univariate partial least squares (PLS) is a method of modeling relationships between Y variable and other explanatory variables. It may be used with any number variables, even far more than the observations. A simple interpretation given that shows to straightforward reasonable way forming prediction equations. Its relationship multivariate PLS, in which there are two or examined, an example it compared by simulation methods With univariate linear combinations variables formed sequentially related ordinary regression. shown these combinations, here called components, viewed as weighted averages predictors, where each predictor holds residual information not contained earlier quantity predicted vecto...