COMPARISON BETWEEN STANDARD PLS, RLS, AND UNBIASED PLS

摘要

Partial Least Squares algorithm (PLS) is widely used in chemometric studies and increasingly in process engineering applications. For process problems, PLS is typically used as a dynamic modeling technique. Unfortunately, traditional PLS identification techniques will typically produce a model that is biased in its regression parameters. In this article, an effective unbiased recursive PLS algorithm is proposed to address this problem. This paper provides a comparative study, using simulated data, to illustrate the potential benefits of the proposed approach.