Abstract (EN):
An approach for doing PLS on very wide datasets is proposed in this work. The method is based on the decomposition, by means of a SVD, of non-superimposed segments of the original data matrix. It is shown that this approach uses less computer resources compared to SIMPLS and PCT-PLS1. Furthermore, it is also shown that the results obtained by this approach are the same as those obtained by other regression methods (PLS and SIMPLS). The method implementation is simple and can be done in a distributed environment.
Language:
English
Type (Professor's evaluation):
Scientific
No. of pages:
10