analyzing data collected from the box (containing accelerometer + gyro sesonrs)
The bach script preprocessAll.sh calls the python script, i.e. the preprocessing pipeline preprocess.py that currently does roughly this:
- cleans telespor export for lat, lon, and batteryvoltage
- calculates speed and engine state (on/off) from telespor data
- resmaples the data to a sampling rate that could be set via arguments and interpolates missing lat lon, and batteryvoltage values
-
adds absolute timestmps to the data using the recorded time of installation
-
resamples (downsamples) the data to the same sampling rate as telespor measurements by averaging meaurements inside the bins
-
For preparing training data, preprocess.py also accepts two timestamps (signalstart & signalend) that limits the measurements used to the time range during which the boat was actually moving
- Features: box measurements (ax, ay, az, gx, gy, gz)
- Labels: engine state (ON/OFF) or batteryvoltage (passed as a sys arg to classify.py)