I am a bit new to classification and ML. I have a dataset that I'm not sure how to handle in regards to the sampling rate of the sensors I am working with, and the time length that each 'event' lasts.

For example, I have 3 sensors that all gather data on 1 event.

Sensor 1 samples at around each .25 second (but it is not exact)

Sensor 2 samples at around each .0125 seconds (again, not exact)

Sensor 3 samples at around each second (not exact).

Each event may be a variable time length (i.e. 2 seconds to 16 seconds).

Currently I am treating each sensor reading of an event as one training example (i.e. For one event, sensor 1's reading and data are fed in as a separate training examples than sensor 2's reading and data)

As above, sensor 2 has a lot more data points per training example than sensor 1. I have time stamps for each reading, but since they don't line up smoothly with the rest of the sensors, I am not sure how to handle it. I am also unsure how to handle it over multiple events as well.

I had considered zero padding the data to just make them all the same length when feeding them into an LSTM, but from what I could see, this was advised against, so I am unsure of where to go from there.

Thanks in advance!


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.