This is not a somewhat ‘correct answer’ - it is more of a workaround.
I had the same issue, the thing that solved the problem for me was setting a constant timestamp difference (e.x 26, so the timestamp would go as follows: 0, 26, 52, …).
Luckily for me, the sensor readings were almost always coming with the same timestamp difference - so it did not affect my data. I hope you will find the ‘correct’ solution to your problem or use this workaround.
Could you share your csv dataset so I can have a deeper look please?
As @Orzech said, we need to have a constant frequency and also I think we only take integers for the timestamp so maybe if you have something like 0, 15.5, 31, … we would only take 0, 15, 31, … so it won’t be constant to our ingestion api in this case. But maybe there is something else?
The sampling rate of the project is 52Hz, so I assumed the time interval is every 19ms. I’m trying to load the file for anomaly and classification testing, yet, it prompt the same error. See the image below.
Ok, I had a quick look at your project and it appears the frequency of all your data is 52Hz but the one you are trying to test (AnomalySampleCSV) is 52.63… Hz thus there is an inconsistency in the frequency: