0

I am currently using scikit-learn to create and fit a SVC classifier. To do this, I wrote the following code:

clf = SVC(probability=True).fit(X_labeled, Y_labeled)

I then use an active learning technique to get more data to improve the classifier's performance. To train the classifier on the new data, can I do this:

clf.fit(newData, newY)

or do I need to do this:

clf.fit(newData+X_labeled, newY+Y_labeled) #pseudocode
jdoe
  • 634
  • 5
  • 19
  • 2
    Yes there is a possibility without having to pass your old data again. It's called online learning. – cs95 Oct 23 '17 at 00:52
  • 1
    I believe that the dupe should address your questions. Best of luck. – cs95 Oct 23 '17 at 00:53
  • From the scikit-learn docs: [Strategies to scale computationally](http://scikit-learn.org/stable/modules/scaling_strategies.html#incremental-learning) – blacksite Oct 23 '17 at 00:54

0 Answers0