First of all, I have begun researching as a masters student and I am forcing myself on more statistical validation and significance testing – this is not my strong suit. As such, I am applying techniques to my current job.
Here is my problem: I have a massive set of data for tested silicon IPs. There is one data measurement that determines the pass fail criteria of the tested parts. However, there are also hundreds of other data measurements taken for the parts. What I was thinking was to create distributions for each measurement, and then perform a PCA on each measurent.
So a more low level description:
I have data pertaining to RxBIST, analog voltages, register status, thermal measurements, etc. The pass fail criteria is determined by the RxBist measurements. Now what I want is to determine what out of status measurements, anaolg voltage measurements, etc., may be causing this failure. My immediate guess would be a principle component analysis.
If anyone has suggestions or can point me to some literature that would be wonderful!
Get help from others!