2

I have a data set of 60 sensors. I wish to decrease the number of sensors used during an experiment, and use the remaining sensor data to predict the removed sensors.

If I were to run principal component analysis on the full data set, would the features (sensors) that correlate with principal component 1 (PC1) recapitulate the rest of the data most strongly, thus highlighting sensors that have the worst correlation with PC1 as those that should be removed (and predicted by the remaining sensors)?

Alexander Fokas
  • 21
  • Unfortunately, I have not enough reputation to post a comment, so i will lose the last :). I've just answer on a very similar question here. Generally, I think You may try to find 1,2,3 N significant principal components and You may find variables (sensors) which you may remove without losing a lot of information. – zlon Jan 25 '17 at 19:29

0 Answers0