CODEWITHSUNDEEP
javaclassificationweka
Sangeeta
Sangeeta

Reputation: 599

WEKA on large training dataset (java code)

I am using WEKA for classification. I am using WEKA jar file in my Java code. My training datset have 56000 instances and 1253 features. I am using simple classification algorithms like logistic regression, J48, Random Forest Random Tree etc. I am able to get results only for Random Forest. While buidling training model for other classifiers my program kept running for 6-7 hours but I did not get any result or error.

Can any one please suggest some faster way to train the WEKA classifiers on large dataset.

Thanks!

Upvotes: 0

Views: 748

Answers (1)

drp
drp

Reputation: 340

As weka's implementation of machine learning algorithms loads entire dataset in main memory it may take huge time to train models sometime. Using Weka via commandline may help in this regard. Also there are some other ways through which training time can be reduced. They are enlisted at http://www.cs.waikato.ac.nz/ml/weka/bigdata.html

Hope it helped.

Upvotes: 3

Related Questions