CODEWITHSUNDEEP
wekanaivebayes
Hassam
Hassam

Reputation: 79

Weka classifier does not run for large number of instances

I am using my data to train Weka NaiveBayes classifier. I am running cross validation. When i run it for 5000 instances, it runs just fine but when I increase the number of instances to 100000, it just stays there showing:

'Building model on training data'

There is no result and weka stops working on the data after a few seconds.

Upvotes: 0

Views: 1169

Answers (2)

gamerx
gamerx

Reputation: 579

The java vm has probably ran out of memory. Increase the heap size in RunWeka.ini to get around this or maybe use less data. You can open and keep open the CLI window before starting training so that you can see the error output.

Upvotes: 0

Pandit
Pandit

Reputation: 748

First, Please share your code. are you using java api or running through GUI. What is the size of the JVM.

The possible reason could be

The textual output shown in the Explorer is that produced when learning the model on all the available data. This is also the model that is saved when you export the serialized classifier to disk. The rationale is that if you were to put a model into production to predict future instances you'd want to use the model built from all the available training data. If you are not interested in this model you can turn this off (and save time) in the "More options" dialog.

provided here

Upvotes: 1

Related Questions