Open IanMayo opened 6 years ago
Thanks! At this moment just one question (to avoid misunderstandings). When you say "steady course" - is it 'constant' bearing/azimuth only? Or maybe, 'constant' bearing AND 'constant' speed?
Unfortunately we can't use constant as a test, since there's wobble in the data.
By eye we can take a pretty good guess what is a turn, and what is just autopilot corrections.
It's that "by eye" that we're trying to automate.
Attached is a .rep file. It has been renamed to .txt so that it can be viewed in GitHub. It should be renamed before use. OtherOwnship_Trimmed.txt
Note: the columns are in the REP format, as recorded here: http://www.debrief.info/tutorial/index.html#reference.html#replay_track_format
I only expect you to read in the date/time, course and speed columns.
The file contains this data:
Our challenge is to determine the periods when the ship is on steady course. I've had a go at marking the straight legs with orange lines.
I'd like you to work on a standalone app that reads in the datafile, and exports a series of start/stop times for the legs that it has spotted.
I'd like you to approach the code in a modular fashion, so we can drop in/out different versions of the slicing code.
Initially, you are welcome to try any type of mathematical/statistical algorithm to do this. We can talk through the algorithms before you start coding, if you wish.