Hyperspectral fitting using hyperspy

Update:

This version includes updates to use weighted PCA in heteroscedastic noise estimation.

This set of Jupyter notebooks are used in the analysis of hyperspectral datasets with the hyperspy library.

The file "clmap_type-IIB_HPHT_diamond-spikes_removed.hspy" is the example dataset.

Summary of the points covered in the different notebooks:

• 01 - Data Cleaning : covers how to fit and remove background, correct imaging shift using cross correlation
• 02 and 03 - Noise Analysis and eV conversion: covers how to measure and model heteroscedastic noise in the dataset using Principal Component Analysis. Conversion from nm to eV spectral representation with jacobian conversion, setting of a variance model in the hyperspy object
• 04 - Model Fitting: Defining and fitting a model on the data. Saving the results
• 04bis - Fit results plot formatting: Formatting plots with all fit results as well as \(\chi^2 \) statistical goodness-of-fit indicators

At the time of initial upload (30.11.2021), hyperspy is in v 1.6.5 which does not support non-uniform axes for data treatment in eV space. The file "Readme - Installing hyperspy from source" contains instructions on how to install the development version.