Using sklearn's GridSearchCV

grid_cv.py

from sklearn.model_selection import GridSearchCV
from sklearn.metrics import make_scorer 
# import Random Forest

param_grid = {'n_estimators': [50, 100, 200],
                 'max_depth':[None, 10, 15, 20], 
                 'criterion': ['gini', 'entropy'],
                 'min_impurity_decrease': [0, 1e7, 1e5]}
scorer = make_scorer(accuracy_score)

clf_grid = GridSearchCV(RandomForestClassifier(), param_grid, scoring=scorer, n_jobs=-1, 
                      return_train_score=True, cv=3)

clf_grid.fit(train_X, train_y)

# to retrieve the best estimator:
clf_best = clf_grid.best_estimator_

# if you want to examine the results:
grid_res_df = pd.DataFrame.from_dict(clf_grid.cv_results_)
grid_res_df.sort_values(by='mean_test_score', ascending=False, inplace=True)
  # then for example:
grid_res_df[['mean_test_score', 'mean_train_score']].head()