It is recommended to follow this procedure : The record process is:
Seek PEC Index
Start guiding
Start PEC recording
Wait until it has completed.
Further PEC records can be done, collecting multiple PEC record cycles and averaging them can produce a better PEC record.
Thus I recorded 3 consecutive PEC files to be averaged.
However these files are so much different that their average would be meaningless.
I attach the content of these 3 files
Has someone got some experience in averaging these .csv files ?
Those first two graphs do not look like a PEC training run. The third one does, but I'll note that it doesn't compare favorably with your guide results (analyzed from your other/separate thread). I would recommend you use Celestron's CPWI and PECTool for this (on your Celestron mount). Both are free, and PECTool does one or multiple runs (batch mode) and auto averages them into the final file for you to upload to the mount. It will also upload to the mount.
edit: Hadn't looked at the INDI driver you referenced. Interesting, and in principle should work. However, I would trust PECTool first, and then cross-check against the INDI driver afterward. From what I see in your 3rd graph, the number of seconds (X axis) does not appear to be right. I could believe the Y axis (roughly matches up with my Celestron mount). For a good average, I'd recommend 5-10 runs. All should look similar to your #3 graph. The major period, magnitude, and small "bumps" on the major period should all be fairly similar in the set. If not, I'd start over and do again until they do reasonably match. Do start with a fairly decent PA...