What does ETA represent in GridSearchCV ScikitLearn?

Question

I am working on a grid search that will have 135 candidates, totalling 405 fits.

Questions:

• What does ETA represent? Does it mean estimated time for the complete grid search to run? I assume so... but I want to make sure as the documentation does not make this clear.

• Also, WHERE do I find mappings for the verbose kwarg in http://scikit-learn.org/stable/modules/generated/sklearn.model_selection.GridSearchCV.html I noticed this in Keras as well.. and I am not sure how we are supposed to know what integer values map to what level of verbosity. From this post, 'verbose' argument in scikit-learn, it seems like one would need to dig through the source. Is that the only way to determine verbosity mappings?

Answer 1

If you look at the implementation of GridSearchCV you can see it creates the object Parallel. The comments inside the implementation of Parallel explain how verbose parameter has been used:

'''
verbose: int, optional
        The verbosity level: if non zero, progress messages are
        printed. Above 50, the output is sent to stdout.
        The frequency of the messages increases with the verbosity level.
        If it more than 10, all iterations are reported.
'''

Since you are using a batch_size=5, ETA says the time that each batch needs.

Basically, GridSearchCV documentation explains that by using verbose parameter you can choose whether visualize the log or not. Then, if you increase verbose you get an higher "refreshing rate" of your training process.