3

I conducted an experiment with multiple human participants to analyze some air traffic scenarios. Some data has turned out to be very unrealistic. Take a look at the following pics which shows the time trace of aircraft positions. Each participant is given the responsibility to fly one aircraft (one time trace line indicates one participant).

This is realistic scenario This one is completely unrealistic

The first one shows a realistic way of flying an aircraft when encountering an obstacle (not the only way, but one of the possible ways). The second one shows a completely unrealistic situation where the aircraft is snaking around. I need to impute this bad data since I can't conduct my experiments again.

My initial approach was to use data from another group for this particular aircraft. However, this will result in having a perfect correlation between the two groups for the particular aircraft.

Any ideas on how I can approach this in a justifiable manner is welcome. Thanks in advance.

tegginamaniss
  • 131
  • 1
    The straightforward thing to do would be to just cast aside the bad observations. Unless there are aspects of the bad observations that aren't bad, that you're perhaps not picturing? – generic_user Oct 27 '17 at 13:45
  • I agree with generic_user. I guess that you actually want to replace these sort of outlying aircraft in an automatic fashion? – keepAlive Oct 27 '17 at 14:17
  • I cannot remove only bad aspects of a given path because this would mean that the aircraft is popping up out of nowhere in some regions. So either I remove it completely, or replace it completely.

    Also there will be some manoeuvres as a response to the bad aircraft manoeuvres. So I think removing bad data completely will result in unexplained manoeuvres by other aircraft. The effect of absence of bad aircraft data on other aircraft still remains the same.

    Though there are good aspects, the fact that this is not a realistic representation makes other aspects lose it's weight.

     – tegginamaniss Oct 28 '17 at 16:21

0 Answers0