I have a 10,000 dimensioned dataset where all attributes are numeric values. I would like to select the best e.g. 50 attributes out of 10,000 so that I can run regression algorithms on it.
I've tried Weka's PCA and CfsSubsetEval algorithms, however they were not capable of handling that much dimension (algorithms never terminated in both case).
What kind of attribute selection algorithms (preferably with already implemented tools) exist in machine learning literature for such high dimensions?
Summary of what I am doing is: I have a log data where all attributes are numeric (10K attributes in total and 100,000 samples in total) and I want to predict the bandwidth out of it. So I want to select best attributes which can play a role in predicting this bandwidth attribute.
