WEKA: Can i combine two machine learning trained models into one model?

Viewed 592

I want to merge two machine learning models which are trained with two different data sets. How can I merge those two models into one instead of building a model by combining those two data sets using weka java library?

Usage: I'm splitting my whole data set (12 million) across the cluster and building individual models (to decrease the training time). So I want to get finally one single model by combining those all models. Is it possible?

1 Answers

You can combine multiple classifiers by Vote classifier.

If you want to code it yourself do something like:

double prediction1 = classifier1.classifyInstance(ins);
double prediction2 = classifier2.classifyInstance(ins);

// use your logic for combining predictions
double combinedPrediction = combinePredictions(prediction1, prediction2); 

Also check https://machinelearningmastery.com/use-ensemble-machine-learning-algorithms-weka/

Related