ZHENQI SHI (shizhenq)
New member Username: shizhenq
Post Number: 2 Registered: 3-2010
| Posted on Sunday, August 05, 2012 - 5:57 pm: | |
I recently came across a request to conduct a head-to-head comparison between MCR and CLS algorithms regarding its capability to deconvolute intermediate or transient species. The CLS algorithms here refer to a series of derivation of the original algorithm, such as prediction augmented CLS, CLS-PLS hybrid algorithm, etc. The sample dataset is an IR spectral set describing a dynamic chemical reaction. After initial investigations, I made following observations. 1. After feeding pure comp spectra to both algorithms (initial guess for MCR and estimated K matrices for CLS), MCR seems to outperform CLS regarding its capability of reconstructing the dynamic reaction spectra. 2. After performing PCA on residual spectra (between original and model reconstructed spectra), feeding PCA loadings to both algorithms was indeed to improve CLS's capability on capturing those potential transient species, while did not help at all on MCR. I am wondering whether there are any prior experiences or similar echos on such a comparison. Any of your comments and questions are warmly welcomed. Thank you in advance. |