The key takeaways from this section include:
- Machine Learning Pipelines create a nice workflow to combine data manipulations, preprocessing, and modeling
- Machine Learning Pipelines can be used along with grid search to evaluate several parameter settings
- Grid search can considerably blow up computation time when computing for several parameters along with cross-validation
- Some models are very sensitive to hyperparameter changes, so they should be chosen with care, and even with big grids a good outcome isn't always guaranteed