As introduced in the previous chapters, apart from the times when using special packages such as XGBoost for extreme gradient boosting and Keras for deep learning, the Python package for machine learning having the lion's share is Scikit-learn.
The motivations for using this open source package, developed at Inria, the French Institute for Research in Computer Science and Automation (https://www.inria.fr/en/), are multiple. It is worthwhile at this point to mention the most important reasons for using Scikit-learn for the success of your data science project:
A consistent API (
fit
,predict
,transform
, andpartial_fit
) across models that naturally helps to correctly implement data science procedures working on data organized in NumPy arraysA complete selection of well-tested and scalable classical models for machine learning, offering many out-of-core implementations for learning from data that won't fit in your RAM memory
A steady development with many new additions...