ML Pipeline for Modeling and Prediction

In the last section, we introduced the concept of the estimator, which chains together different transformation processes to make feature extraction easier. We also saw the demonstration of the fit and transform functions for the estimator we built. Estimators have far more capabilities than just fit and transform. Estimators can also be used to chain together classifiers such as logistic regression, KNN, or random forest classifiers along with the transformation steps.

When classifiers are introduced into the estimator, the estimator also inherits many of the functions of the classifiers, such as scoring and predicting.

So, now we have a single engine that is capable of performing very diverse functions that otherwise would have to be performed by separate functions. Here lies the beauty of the pipeline utility, which enables us to build all-encompassing functions in a single engine.

In the next exercise, we will demonstrate the...