This question already has an answer here:

I have a large $(10000 \times 5001)$ table representing $10000$ samples and $5001$ different features of these samples. One of these features represents an output variable of each sample. In other words, I have $5000$ input variables and one output variable for each sample.

I know that most of these inputs are irrelevant. Therefore, what I would like to do is determine the subset of input variables that predicts the output variable best. What is the best/simplest way to go about doing this in R?


marked as duplicate by steffen, gung, whuber Apr 30 '13 at 14:38

This question has been asked before and already has an answer. If those answers do not fully address your question, please ask a new question.

  • $\begingroup$ to all reviewers: Please see that the answers in the duplicate question are NOT focused on binary classification, but can be applied to multiclass or regression problems as well. $\endgroup$ – steffen Apr 30 '13 at 9:05

What people typically do is test the correlation between each feature and the response compute, save and order the p-values and then drop everything but a small percentage with the lowest p-values. Don't take the p-values seriously. This was just intended to be a quick screening device. Once you are down to a small enough number you can go ahead with the standard variable selection techniques used in regression.

  • $\begingroup$ And could you please say what standard variable selection techniques are? $\endgroup$ – Olga Feb 7 '13 at 15:55

Not the answer you're looking for? Browse other questions tagged or ask your own question.