GridSearchCV for my Random Forest breaks up. I need to know the reason and the solution to make it work:

# Grid-Search for Random Forest
param_grid = {
    'bootstrap': [True],
    'n_estimators': [100, 200, 300, 400, 500],
    'max_depth': [50, 100, None],
    'max_features': ['auto', 200],
    'min_samples_split': [2, 5],
    'min_samples_leaf': [2, 5],
    'oob_score': [True],
    'warm_start': [True]    

# Base-Model for improvement
rf_gridsearch = RandomForestRegressor(random_state=42)

# Grid-Search initiation 
rf_gridsearch = GridSearchCV(estimator = rf_gridsearch, param_grid = param_grid, 
                           scoring = 'neg_mean_absolute_error', cv = 5, 
                           n_jobs = -1, verbose = 5)

# Perform the grid search for the model
rf_gridsearch.fit(X_train, y_train)
  • $\begingroup$ What do you mean by "breaks up"? An error message (provide it please), python crashes, your computer crashes, ...? $\endgroup$ Jan 24 '20 at 13:46

First, you are fitting $5 \cdot 3\cdot2\cdot2\cdot2\cdot5=600$ models and n_estimator=500 is quite big. Of course, this depends on your dataset and in your computing power.

My first guess will be that you have not enough RAM memory on your laptop(if you are running it there) and that is why it is collapsing.

If the error is this one, I recommend sampling your data to 1/10 or less (depending on your data) and searching for the best hyperparameter there and then using your whole data for the final model.

  • $\begingroup$ @ml_learner I have noticed that in your questions you never mark the accepted answer. When someone answers your question you can do 2 stuff: upvote (up arrow) and mark the answer as correct (green tick) Check this link meta.stackexchange.com/questions/5234/… This helps other users know what is the correct answer and give the person who helped you a bonus $\endgroup$ Jan 25 '20 at 9:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.