Larch for XAFS Analysis: 4 Linear Combination and PCA XANES Analysis

[@Frh0412]

Thank you for uploading very useful lecture. I have a question about the goodness of LCF fits. It would be nice if the fitting result and data are well matched, sometimes it looks to be pretty matched but there are several mismatches. I wondering how could I decide the fitting results are matched or not. Is there any rule of thumb to determine whether the fitting is fitted or not? Actually I usually use Athena software and it gives me several statistical parameters such as R factor, chi square, and reduced chi square. Are they can be used for that?

[@MatthewNewville]

When you do a linear combination fit with XAS Viewer, the results are shown in a window titled "Linear Combination Results". That will show all the linear combinations for a particular group, sorted in order of reduced chi-square. Fit statistics (n_vary, chi-square, reduced chi-square, R-factor, Akaike Info criteria) are shown for each fit in the middle portion of this window. Whether the fit is actually "good" or not will depend on whether to components used as standards actually model the unknown spectrum (or spectra). Generally, an R-factor below 0.01 (that is, a 1% mismatch) would be considered "pretty good", and whether it is good enough for your needs depends on what you're trying to (and able to) get out of your data. The usual effects that make XANES analysis challenging (consistent normalization, energy alignment, pinhole effects, overabsorption effects) will affect any linear combination analysis, but these can be minimized with some care in both the experiment and the data processing.

[@Frh0412]

Thank you. It helped me a lot!

[@ardalanhayatifar4034]

Thanks for the videos. Does Larch have the feature to show SPOIL values after target transformation?