Optimization in Locally Weighted Regression

摘要: The application of locally weighted regression (LWR) to nonlinear calibration problems and strongly clustered data often yields more reliable predictions than global linear models. This study compares the performance LWR that uses PCR PLS regression, Euclidean Mahalanobis distance as a measure, uniform cubic weighting objects in local Recommendations are given on how apply near-infrared sets without spending too much time optimization phase.