astroML

AstroML, initiated in 2012 by Jake Vanderplas, Željko Ivezić, Andrew Connolly, and Alex Gray and now maintained by the AstroML developer community, is a Python library providing statistical and machine learning routines specifically for analyzing astronomical data. It's built on NumPy, SciPy, scikit-learn, matplotlib, and Astropy, ships with free astronomical dataset loaders, and includes extensive examples covering clustering, dimensionality reduction, regression, classification, and time-series analysis via Jupyter notebooks and a visualization gallery. A second edition of the accompanying textbook was published in 2019 with updated code, and the project maintains active errata documentation. Completely free and open source under the 3-clause BSD license. Good fit for an astronomer or astrophysics researcher who wants machine learning tools built specifically around real astronomical datasets and use cases, rather than adapting general-purpose ML libraries.