NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: On checking accuracy
From: Bill B
Date: 2008 Aug 04, 17:48 -0400
From: Bill B
Date: 2008 Aug 04, 17:48 -0400
Peter wrote > Its a nice explanation and I recommend it. However, its not the > article on the Starpath site that I think I remember (I know ..). > THAT one (in my wild erratic fancy) showed how an outlier can lead to > quite a wrong slope, if all you had to work with was the data points > themselves. Bill, does this ring any bells? Anyone? Thanks for digging up the link. It can see it my minds eye, but I cannot factually verify it. I have performed a comparison on some of my data, Hc slope vs linear-regression trendline (computed by Excel) and found the slopes to be different, and outliers falsely identified with the trendline. That may be the mental image I have. Whether I ever posted the results I do not know. Burch does use the same data in both the graphs (first best fit/trendline, second best fit to Hc slope). He DOES point out in the text that what looked like an outlier ( 2nd observation) in the poor-man's linear regression (graph 1) was indeed not so, which a fit to the actual slope made this clear (graph 2). It would be nice if he used both methods on the same graph, so it would be clear to the casual observer. Bill B. --~--~---------~--~----~------------~-------~--~----~ Navigation List archive: www.fer3.com/arc To post, email NavList@fer3.com To , email NavList-@fer3.com -~----------~----~----~----~------~----~------~--~---