If you have the time and computational resources, then you would want to try hyperparameter tuning with many different models before selection one. However, for this project, I wanted to keep it manageable so I used the default hyperparameters for model selection.

There are some cool projects using automated machine learning to find the ideal model and hyperparameters given a dataset. H2O has an auto-ml framework as well as auto-sklearn. Automated model selection is probably the future of data science, so it’s worth exploring these options.