combine py-earth with sklearn.ensemble.BaggingRegressor

[zrq0123]

Hi, in my project, I combine pyearth with BaggingRegressor to get a bagging result, my code is as follow:
bagger=BaggingRegressor(base_estimator=Earth(penalty =20,feature_importance_type="rss",max_terms =10,use_fast=True,allow_missing=True),n_estimators=1000,n_jobs=-1,random_state=random.randint(1,100),bootstrap=True,max_features =5)
In one small scale dataset, the code run correctly, but in a large one (4381 cols * 832 rows), the code give a ValueError: assignment destination is read-only, and if I set the parameter n_jobs =1 ,the code can run correctly. So the py-earth have some bugs or error combine with BaggingRegressor on multiple CPUs, which can speed up the computing

[jcrudy]

@zrq0123 Thank you for reporting this. Can you give me the full stack trace for this error? Also, if possible, can you give me detailed code, data, or steps to reproduce the error? I know it can be difficult to do this with data-dependent errors, but anything you can provide might help. Does it happen on every sufficiently large input? Finally, can you give me information on your platform: operating system, python version, version of scikit-learn, scipy, numpy?

[zrq0123]

Thanks for you quick replying, I have sent the code and dataset to your Gmail.

[jcrudy]

This should now be fixed in master. Closing issue. @zrq0123, please comment and reopen if this problem persists for you.