Tuning parameters for a dataset/algorithm (grid search)

[weixuanfu]

• Add a tuning flag
• Save best result as output
• Save other results into result folder
• Add UI/API for this function

[weixuanfu]

• Tuning Flag is --grid_search (default value is False). For turning on parameter tuning, just assign --grid_search True via API, e.g. python driver.py --method LogisticRegression --grid_search True --_id xyz ...
• The GridSearchCV results will save under result folder and its name pattern is grid_search_results_{_id}.csv. This table includes all the parameter combinations and the mean_train_score column is the average CV scores in GridSearchCV. For those invalid parameter combinations, the mean_train_score is '-inf'.