Pipeline

Python scikit-learn provides a Pipeline utility to help automate machine learning workflows.

Very useful for cleanly streamlining…

1. Data Transformation
2. Model Selection (can we pass an algorithm as a parameter?)
3. Hyperparameter Tuning
4. Save Pipeline

Example #1

1. Standardize Data
2. Learn a Linear Discriminant Analysis model

#keynotes: cross_val_score on pipeline is very efficient, but convert your metric as a score with make_scorer().

Example #2

1. Feature Extraction with Principal Component Analysis (3 features)
2. Feature Extraction with Statistical Selection (6 features)
3. Feature Union
4. Learn a Logistic Regression Model

#keynotes: you can use pipeline to transform data with pipeline.fit and pipeline.transform.

Example #3

1. Select Features
2. Imputation by 0
3. Imputation by mean
4. Feature Union

Example #4

1. Repeat Example #2
2. Grid Search on “n_components” and “k”
3. Use “best_estimator” for prediction

Example #5

1. Save pipeline/GridSearchCV
2. Load pipeline/GridSearchCV