added Find_unscaled to figure out which mmsi were unscaled. Turns out to be just 3, all in longliners.
edited vessel_scoring/add_measures to take an unscaled argument. MMSI in unscaled have their speed and
course divided by 10.
edited scripts/add_measures to automatically add the appropriate unscaled argument
Edited to vessel_scoring/add_measures to compute std deviation in a different way that avoids the discontinuity 0/360.**
Number 4 could be pulled out and considered as part of a different branch. That was what I was experimenting on that led me to notice this issue. The idea is to convert the course to cos(course) and sin(course), then take the RMS of the standard deviation of these to values to get a total standard deviation like quantity for course that does not have a discontinuity.
Closes https://github.com/GlobalFishingWatch/vessel-scoring/issues/16
unscaled
argument. MMSI inunscaled
have theirspeed
andcourse
divided by 10.unscaled
argumentNumber 4 could be pulled out and considered as part of a different branch. That was what I was experimenting on that led me to notice this issue. The idea is to convert the
course
tocos(course)
andsin(course)
, then take the RMS of the standard deviation of these to values to get a total standard deviation like quantity for course that does not have a discontinuity.