Scikit-learn

Scikit-learn is a free, open-source machine learning library for the Python language, focused on “classic” machine learning—supervised and unsupervised—and not on deep learning. It is distributed under a BSD license, which allows commercial use, and it is one of the most widely used tools in the field.

What it offers

It covers the main families of classic machine-learning tasks: classification (support vector machines, random forests, gradient boosting, k-nearest neighbors, logistic regression), regression, clustering (k-means, DBSCAN), dimensionality reduction (PCA, NMF), plus model-selection tools (cross-validation, grid search) and preprocessing. It is built on Python's numerical libraries, NumPy and SciPy.

Origin

It began in 2007 as a Google Summer of Code project by David Cournapeau, and in 2010 a team at the French institute INRIA—Fabian Pedregosa, Gaël Varoquaux and others—published the first public release. Its reference paper, “Scikit-learn: Machine Learning in Python,” appeared in the Journal of Machine Learning Research in 2011.

Why it matters (and what it is not)

Its great hallmark is a consistent API: all models share the same pattern—they are fit with fit, predict with predict and transform with transform—which, together with its documentation, has made it the de facto standard tool for classic machine learning and for teaching. It helps to clarify what it is not: it is not a deep-learning framework—for deep neural networks one uses PyTorch or TensorFlow—although it integrates well with them.