shamailahmad875/EELS-Spectrum-Image-Clustering-UMAP-and-MCR-LLM ? reverse-engineered prompt

Build me a Jupyter notebook for unsupervised analysis of EELS spectrum image data, where each pixel has a full energy loss spectrum.

I want two ways to explore the data, first a UMAP based workflow that reduces the spectra and clusters the pixels, and second an MCR LLM workflow that separates the map into non negative component spectra and abundance maps. Please make it easy to load a preprocessed HyperSpy dataset, crop to the energy range I care about, run the analysis, and show clear visualizations of the cluster maps, component maps, and average spectra.

It would be great if the notebook explains the steps in simple terms, includes a few sensible defaults, and saves the results out as files I can open later, including component maps and resolved spectra. If you need current package details or syntax, look up the latest docs online. Keep it practical and focused on getting useful chemical or phase contrast from a big spectrum image without assuming I already know how many components there are.