Memory Explosion with GradientBoostingClassifier, 'mae' criterion, and GridSearchCV #8239
Comments
|
Great - thanks |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
{{ message }}
Description
Memory leak causing system to freeze when using GridSearchCV.fit() with GradientBoostingClassifier with n_jobs > 1, cv=ShuffleSplit, and criterion = 'mae' in the param grid. Change the loss criterion from 'mae' to default and the problem goes away. I am using a tiny dataset (<1000 data points, 11 numeric predictors, binary target).
Steps/Code to Reproduce
from sklearn.ensemble import GradientBoostingClassifier
from sklearn.model_selection import ShuffleSplit
from skleran.model_selection import GridSearchCV
ss = ShuffleSplit(n_splits = 50, test_size = 0.5)
gbc=GradientBoostingClassifier()
param_grid = {'n_estimators': [50], 'learning_rate': [0.05], 'criterion': ['mae'], 'max_depth': [1,2,3,4,5,6,7,8,9,10,15,40]}
clf = GridSearchCV(gbc,param_grid = param_grid, cv=ss,scoring='precision',n_jobs=4,pre_dispatch=4)
clf.fit(predictors, target)
Expected Results
This should just complete running easily with my tiny data set (<100KB).
Actual Results
Instead the parallel threads rapidly grow in size until they take over all 24GB of memory on my machine.
Versions
Windows-7-6.1.7601-SP1
('Python', '2.7.11 |Anaconda 2.5.0 (64 bit)| (default, Jan 29 2016, 14:26:21) [MSC v.1500 64 bit (AMD64)]')
('NumPy', '1.11.1')
('SciPy','0.17.0')
('Scikit-Learn','0.18.1')
The text was updated successfully, but these errors were encountered: