Week 9: Testing the Classifier

Apr 21, 2019

During Week 9, I talked to my external advisor about the progress I have made and he gave me good advice on what I should be doing this week. I began tuning my classifier and trying to find which methods of the classifier get better results. I used methods such as setMem(), which changes the number of iterations the classifier trains the training data, and setEpsilon(), which changes the amount of smoothing of the classifier. With these methods, I got around 70% accuracy using the n-fold cross validation technique with around 200 test articles and around 600 training articles. I could not use all my data, which is 1600 articles, because I kept getting the java.lang.OutOfMemoryError: Java heap space error. This meant that I have to run my program on the command line or terminal to use all the data. Also, I started working on my final presentation.

Next week, I will be trying to figure out how to run my program from the command line and keep working on my final presentation.

Thanks for reading!

Leave a Reply

Your email address will not be published. Required fields are marked *