Comments on Fitting a rational function in R using ordinary least-squares regressionTypePad2017-03-23T21:55:42ZBlog Administratorhttps://blog.revolutionanalytics.com/tag:typepad.com,2003:https://blog.revolutionanalytics.com/2017/04/fitting-rational-functions-with-lm/comments/atom.xml/Ben Bolker commented on 'Fitting a rational function in R using ordinary least-squares regression'tag:typepad.com,2003:6a010534b1db25970b01b7c8eaf43a970b2017-04-07T21:46:37Z2017-04-21T15:00:00ZBen BolkerAgree that this is cool. If I'm not mistaken, this works nicely for getting the expected relationship but will mess...<p>Agree that this is cool. If I'm not mistaken, this works nicely for getting the expected relationship but will mess up inference (because the assumptions about independence, Gaussianity, etc. no longer hold), by analogy with older methods of fitting hyperbolic functions as 1/y ~ 1/x ?</p>Carl Witthoft commented on 'Fitting a rational function in R using ordinary least-squares regression'tag:typepad.com,2003:6a010534b1db25970b01b7c8ea7ba6970b2017-04-06T18:02:51Z2017-04-06T18:17:49ZCarl WitthoftNice methodology. I'll just point out that this only works with single-valued functions. I've been playing with algorithms to fit...<p>Nice methodology. I'll just point out that this only works with single-valued functions. I've been playing with algorithms to fit conic sections, and they are quite a different kind of mess!</p>