Most machine learning problems we have discussed so far consist of at least a preprocessing step and a classification step. The more complicated the problem, the longer this processing chain might get. One convenient way to glue multiple processing steps together and even use them in grid search is by using the Pipeline class from scikit-learn.