Deciding what to try next in Machine Learning

Deciding what to try next in Machine Learning

Possible choices:
Get more training examples
Try smaller set of features
Try getting additional features
Try adding polynomial features
Try increasing \(lambda\) (regularization parameter)
Try decreasing \(lambda\)

Each one of those avenues takes time to see if they're promising. By using Machine learning diagnostic will give us insight what is/isn't working with a learning algorithm. It can take time to implement diagnostics, but it will well worth the time and effort in the long run.