update trained model

[kengz]

currently calling

classifier.fit(X_train1, y_train1)
classifier.fit(X_train2, y_train2)

resets the classifier and train it on training set 2. Is there a way to "resume training"?

Example use case would be getting the next batch of training data next, or do on-line training, e.g. incoming new labeled data while the model is deployed.

[terrytangyuan]

There's a continue_training parameter you can use. Take a look at some examples, in particular this one.

[kengz]

exactly what I'm looking for. Thanks!