Scikit-learn, A Python Machine Learning Library, Gets New Feature Updates in Version 1.0


After a long 14 years of development, the latest version-1.0 has been announced for scikit-learn, a Python machine learning library, with major updates and fixes that will allow it to scale even further with sufficient stability.

In fact, version 1.0 of scikit-learn focuses mainly on the stability to make common user functions more robust. The scikit-learn library is built on top of NumPy, SciPy and Matplotlib. It supports supervised or unsupervised learning with various tools for model fitting, including data preprocessing and evaluation available in the toolbox. This new version of scikit-learn requires python 3.7+, numpy 1.14.6+ and scipy 1.1.0+. More details on required sets can be found here.

The first important update to scikit-learn is keyword parameters. The official explanation for this addition in the API identifies that too many inputting of data can confuse users, so it’s more straightforward with just one or two values needed when performing certain functions.

To improve the readability of scikit-learn code, users need now to provide names for most parameters as keyword arguments instead of positional ones. Positional argument passing has been deprecated from version 0.23 and will result in type errors if misused.

An online one-class SVM implementation that uses stochastic gradient descent has been added to the Scikit-Learn library. This could be useful for cases where large numbers of training samples are needed, such as fitting linear classifiers.



Other Source: