|
Hi all, I have a very strange thing happening here. I have an arff file containing roughly 20k instances as data set. With this file, when I run logistic classifier, it takes forever building the model (I waited kind of 20 mins to finish it). But when I save this arff file as csv file, and load the csv file into weka and run the logisitic regression, it takes 4 seconds building the model and doing the stuff as expected. What is going wrong with this? is it equivalent to use csv file instead of arff file? Obviously the csv file does not contain the meta data. Is it harmful for my classification process? I can't explain why is happening. any idea or experience with it? |
