I have a set of time series sensor measurements (acceleration and gyroscope readings) for driving events (harsh acceleration, harsh brake …) with the type, start and end of each event. I need to classify those events in real time and the problem is that each event has different window’s size. What are the lachine learning techniques that I can used ?

  • $\begingroup$ Do they have the same frequency (minute data, hourly data, etc...?) $\endgroup$ – Skander H. Mar 22 '19 at 20:54
  • $\begingroup$ Yes, 50 samples per second $\endgroup$ – Tatou Mar 25 '19 at 10:35