Open mathause opened 3 weeks ago
I thought about this too but in the paper the term harmonic model refers to the combination of the linear regression with yearly temperature and the fourier series for the seasonal cycle. So I did not want to divide it up in the main code. We could make fit_harmonic_model
call the linear regression and the fourier series separately but as we don't actually do the linear regression this doesn't make much sense.
Of course going forward we could drop the name harmonic model and just have a Fourier series around the yearly temperature but what I wrote above is the reason why its not already like that.
Going through the MESMER-M example notebook (#491) I found it non-obvious
mesmer.stats.fit_harmonic_model
also removes the annual mean from the monthly data. Now that this is no longer estimated - should we add dedicated functions to add/ remove the annual mean?