Notes from the Matflow lab
Practical write-ups on instrument data, sparse-data modeling and formulation screening — written the same way the platform works: with the limits stated.
Published notes
Each post ends with a working path you can run — in the product or in your own Python environment.
Most solubility tools give you three numbers and false confidence. Matflow's Hansen HSP model reports intervals and an applicability-domain flag — here is how to use it in five minutes, and when it fails.
Neware .ndax exports hide degradation mechanisms behind cycle counts. Here is how Matflow fits degradation modes, picks by held-out RMSE, and reports remaining useful life with split-conformal intervals.
Bio-Logic .mpr files drift between EC-Lab versions and hide channel metadata in binary headers. Here is a Python (galvani) route that works, plus a no-code path that returns a governed dataset.
Read it, then run it
Every method in these posts has a free-tier path in the product: upload a file, review the parsed rows, and export a governed dataset.