NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: Rejecting outliers
From: Peter Fogg
Date: 2011 Jan 1, 17:02 +1100
From: Peter Fogg
Date: 2011 Jan 1, 17:02 +1100
Peter Hakel wrote:
Its not just a question of competency, the conditions of observation have a lot to do with how good or bad a round of sights is going to be. Poor horizon, the difficulties of measuring an altitude accurately on a platform that is moving in several directions at the same time, etc.
Average to good conditions out at sea means a swell of a couple of metres. This is nothing for a biggish ship and rightly considered as relatively calm conditions, but the same is not the case for a small boat. And often the horizon is less than ideally clear. It takes some practice to get reasonable sights while underway. Producing a really good series of sights of the same body over a short period, like the ones that are occasionally displayed here, is not likely unless the sea is unusually calm. If only what you propose was true. It would make the task of analysis relatively simple.
Recently the focus seems to have been on whether outliers should be removed from a population or not, and if so within which parameters, and so on. I really don't have much to contribute to this discussion as my approach is more pragmatic. Often the GPS is functioning as it should, so the slope (and consequent bisection of intersecting position lines fix) can be compared with it. Use of slope does quite often result in the adoption of a better altitude/time combination than any of the individual sights. Slope does work. If it didn't then I wouldn't be interested in it.
But this question about how to place the slope amidst the pattern of sights - this is how slope works, remember, its not number crunching - is up to the individual navigator. Its necessarily intuitive to some extent, or at least it is for me, assisted by my memory of how good or not I felt about an individual sight at the time I recorded it - whether or not the boat was relatively calm at that moment, whether I was measuring from the top of a wave, and so on.
Practice helps.
... I think we can assume that the vast majority of data collected by a competent navigator will be good and thus the removal of the rare outliers is justifiable.
Its not just a question of competency, the conditions of observation have a lot to do with how good or bad a round of sights is going to be. Poor horizon, the difficulties of measuring an altitude accurately on a platform that is moving in several directions at the same time, etc.
Average to good conditions out at sea means a swell of a couple of metres. This is nothing for a biggish ship and rightly considered as relatively calm conditions, but the same is not the case for a small boat. And often the horizon is less than ideally clear. It takes some practice to get reasonable sights while underway. Producing a really good series of sights of the same body over a short period, like the ones that are occasionally displayed here, is not likely unless the sea is unusually calm. If only what you propose was true. It would make the task of analysis relatively simple.
Recently the focus seems to have been on whether outliers should be removed from a population or not, and if so within which parameters, and so on. I really don't have much to contribute to this discussion as my approach is more pragmatic. Often the GPS is functioning as it should, so the slope (and consequent bisection of intersecting position lines fix) can be compared with it. Use of slope does quite often result in the adoption of a better altitude/time combination than any of the individual sights. Slope does work. If it didn't then I wouldn't be interested in it.
But this question about how to place the slope amidst the pattern of sights - this is how slope works, remember, its not number crunching - is up to the individual navigator. Its necessarily intuitive to some extent, or at least it is for me, assisted by my memory of how good or not I felt about an individual sight at the time I recorded it - whether or not the boat was relatively calm at that moment, whether I was measuring from the top of a wave, and so on.
Practice helps.
