Description David Williams 2014-10-21 11:58:25 EDT

While I'm only beginning to learning about the "performance testing" system, I was told recently (bug 443971) that for the various test cases it is "statistics" that are stored in the database, not the "raw data". 

This surprises me, since typically, the raw data would be stored, and then statistics computed in "post processing" stages. 

The reason the raw data is better is then the data could be subject to many additional statistical analyses, such as "multivariate analysis", or other methods, whereas if "means and standard deviations" are stored, it makes these other methods hard, if not impossible, to implement. 

Plus, having the raw data makes it possible to change "policy" on what to do with outliers, etc. 

I'm not at all sure what it would take to implement this, nor if here is time or interest to invest in it, but I didn't want to lose the idea. Especially since this could impact others that make use of the "performance system".