Hi, We are uploading CSV training data that has a sample rate of > 60s
The data is gas concentration values from several different gas sensors, temperature and rh. Certain gas sensors have a long integration time of > 60 seconds to register the full gas concentration change and we cannot sample the data faster. Is it possible to upload this data or do we need to somehow interpolate or rescale the data? Or perhaps we should just get rid of the timestamp? We’re attempting to build a NN classifier.
Hello Aurelian and thank you for the response. The error I get is that “interval_ms should be under 60000” implying that it doesn’t like sample rates greater than 60000 ms (or 60 seconds). Our sample rate is about 123 seconds (or 123000 ms) the deviation between sample rates is < 1 second. The exact output is:
Uploading 1 file...
[1/1] Failed to upload smoking-covered-IRAP1.csv interval_ms should be under 60000
Done. Files uploaded successful: 0. Files that failed to upload: 1.
Job completed
Indeed it is currently set to a max value of 60 sec. I filled a bug report to our studio team.
In the meantime, you can use any timestamp value lower than 60 sec. You’ll need to adjust the window size to something smaller in order to capture the expected number of data points.