The recommended method is to usejoblib, this will result in a much smaller file than a pickle:

from sklearn.externals import joblibjoblib.dump(clf, 'filename.pkl') #then your colleagues can load itclf = joblib.load('filename.pkl')

See theonline docs

Have you tried pickling theRandomForestClassifier using the Pickle module and then saving it to the disk?

Here’s an example based on thepickle docs:

import pickleclassifier = RandomForestClassifier(etc)output = open('classifier.pkl', 'wb')pickle.dump(classifier, output)output.close()

The “other people” could then reload the pickled object as follows:

import picklef = open('classifier.pkl', 'rb')classifier = pickle.load(f)f.close()

• python
• scikit-learn
• random-forest