0

I am working on naive bayes and logistic regression using prediction io, it is great product. All are working fine. but when I build the prediction model and deploy, it gives java heap error so I gave this command pio deploy -- --driver-memory 2G and error removed. I have 4 models and all are taking 2GB. My question is how to optimize the memory uses in prediction io. I have 8GB RAM System and I have to build and deploy 7 different prediction models.

Thanks

Kishore
  • 5,761
  • 5
  • 28
  • 53
  • Are you doing this on one machine? – pferrel Oct 15 '15 at 18:33
  • Yes, will you suggest me optimize way. – Kishore Oct 16 '15 at 04:58
  • PredictionIO is meant to scale vertically with more memory or horizontally with more cluster nodes. On a single machine all you can do is make sure your ids take up as little memory as possible in the driver. Shorter ids will shave a little off memory useage but not a lot unless you have very long ids. – pferrel Oct 17 '15 at 15:24
  • make sure your ids take up as little memory as possible in the driver. I am not getting your point. What is ids here. Can you please little bit explain it. – Kishore Oct 21 '15 at 07:05

0 Answers0