Hello All,
I am using WEKA CLI to classify different .arff files using metacost algorithms as J48, RF, SVM and NB. I have a large data that contains around 1 million .arff files which needs more than 2 months to run on these algorithms.
I searched for different solutions, and one of them states to increase the memory size for Java. I have did this (as illustrated below) but it didn't enhance the performance.
"java -Xmx4096m weka.classifiers.meta.MetaCost -C Cost_10.cost -W weka.classifiers.trees.J48 -t comb_1.arff -x 10"
I am wondering if Python has different libraries for cost sensitive algorithms (for C4.5, RF, NB and SVM) and are faster than Weka.