Re: [Gretl-users] Regression coefficients

On Thu, 7 Jan 2016, Marc O'Callaghan wrote: >> I don't remember exactly, offhand, what the project was here, but it >> sounnds like you need to look at the $coeff accessor (to grab the exact >> coefficients after estimation) and the lincomb() function (to create a >> series via a linear combination of terms). >> >> Allin Cottrell > > Thanks for your answer, though I think I didn't express myself correctly. > For the context : I am studying the evolution of max and min temperatures > throught Feb-March-Apr of every year, with special emphasis on March. My > main goal is to sort them into two categories : 1) when they have a > tendency to rise during the month and 2) when they tend to sink. > So as to avoid plotting a graph for each year and judging on the basis of > that what the tendency is, I was planning to plot the 3rd-degree > regression using the "polynomial trend" (or whatever it's called in > English, I' sorry, I downloaded Gretl in French to better understand what > the others in my group talk about) filter. Can I somehow find out what > that polynom is ? Does this seem like an intelligent way to go ? > If not, could you please be so good as to explain how you would recommend > I proceed, keeping in mind that I am (to my great shame) an utter > nincompoop in matters Gretl ? If your dataset looks something like what I recommended in http://lists.wfu.edu/pipermail/gretl-users/2015-December/011431.html where "yrday" is a series that runs from 1 to 90 for the 90 days starting on Feb 1 each year and TMAX records maximum temperature, then you could run a regression as follows series yd2 = yrday^2 series yd3 = yrday^3 ols TMAX const yrday yd2 yd3 That will give you your 3rd degree polynomial trend for TMAX. Allin Cottrell _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-users