Machine Learning Model Saving Options

Question

Any suggestions / recommendations for preserving and reusing trained machine learning models? I develop models in Python or R. Then these models should be used in the production process for scoring (where R is not available). For example, there may exist a logistic regression model prepared in R. Now new observations must be taken for this model. The scoring engine must be fast and scalable. I was thinking about the following

• PMML ( http://en.wikipedia.org/wiki/Predictive_Model_Markup_Language ). It is easy to convert most models developed in R to pmml. However, I could not find a useful counting mechanism for PMML models. For example, there is augustus ( https://code.google.com/p/augustus/ ), but it implements only 3-4 models.

• Serialize models using pickle in Python and write to the consumer in Python.

Any thoughts / suggestions on the right approach?

Answer 1

Scikit-learn, pickle . , , .

, JSON . . , - Python, JSON .

Answer 2

, pickle python, :

import pickle
s = pickle.dumps(clf)
clf2 = pickle.loads(s)

joblib, , numpy , scikit-learn.

from sklearn.externals import joblib
joblib.dump(clf, 'filename.pkl')
clf = joblib.load('filename.pkl') 

API- RESTful.