averaging multiple results for compare

[hotaruhime]

Hi! I've been searching for a while and haven't found anything around this topic.

For my current use case of the PTS, I would like to run the same test/suite in multiple instances of an AWS flavor. Having all those results, I would like to compare them against the averaging of another flavor.

An example to make it more clear:

I have a test suite (Suite A) with 2 tests.

• I launch 10 "c4.xlarge" instances and run Suite A in all of them. This gives me 10 different results folders/composites/etc. I call this Run A
• I launch 10 "c5.large" instances and run Suite A in all of them. This gives me 10 different results folders/composites/etc. I call this Run B

I don't want to compare the 20 results separately, but to make an average of all the results for Run A (Avg A) and another average of all the results of Run B (Avg B) and then compare Avg A vs Avg B

I've read the documentation and test different options with no luck on finding something to achieve this. So, my question is:

Is there any way within the current solution to achieve this?
If not, could using the raw data of the tests results and averaging those myself give an ""accurate"" value?

Thanks in advance!

[Michael]

Nothing comes to mind for easily achieving that... If I understand correctly, aside from the 'geometric mean' option? If not that, At the moment the likely easiest way would be using the CSV export option and opening up the data in Excel / LibreOffice Calc and manipulating it that way.

[hotaruhime]

Thanks for your prompt response Michael!

If I went that route, would I be able to use the result viewer somehow? I've haven't found a sort of "import" for csv or json.

Thank you