maruedt/chemometrics

Python library for chemometric data analysis

chemometrics's Issues

Adjust `asym_ls` to `numpy.linalg.lstsq` syntax

Allows for drop-in replacement if the function syntax and the return arguments of asym_ls and numpy.linalg.lstsq are the same. Furthermore, some speedup may be possible by reusing the residuals as often those are interesting with asymmetric least-squares regression.

Change sparse matrix linear equation solver in whittacker smoother

whittacker currently relies on the cojugate-gradient method for solving the linear system. scipy.sparse.linalg provides methods for prefactorizing the relevant sparse matrix (LU-factorization) which can further speed up the calculations if whittacker is applied to many samples simultanously.

Use `fit_base` in `fit_pca`

base.fit_base is a function for selecting the optimal number of latent variables in a latent variable model based on a . Currently, base.fit_base is however only used during the call to regression.fit_pls. To reduce code dublication, decomposition.fit_pca should be adjusted to use base.fit_base.