[Gretl-users] Re: [Off-topic] Data scientists vs Statisticians

Excellent reply from Franck Nadaud. Me, as an outsider of this discussion (I have a degree in Operations Research, but I work as Software Tester), I have a neutral position, or more like a grey color in the white vs black discussion (pick one color for Statistics ;) ). I would say the Econometricians would be more on the grey side but with a bias to the Statistics color :). We can see the advantages of Data Science, in processing big samples of data and extracting patterns that a human would not be able to see (or too live long enough to find them). (I am mostly thinking on DNA, medical/medication research). I am not advocating for either side, there are advantages in all these (three sciences), it is a matter of selecting the adequate tool to solve the problem at hand. Currently, there is this feeling, that everyone can be a Data scientist, by just applying a bunch of algorithms and draw conclusions from them. I think that a good background in Statistics is essential to be a Data scientist (as it is for an Econometrician). When I think about this Computer Science vs Mathematical reasoning, I always think about the Four Color Theorem <https://en.wikipedia.org/wiki/Four_color_theorem>which was possible to proof with a computer executing, a mathematical model. These days we need to stop COVIR-9 spreading in the world, hope scientists with whatever science, tools, or techniques, they find good solutions (and political powers listen to them). Thank you all, for keeping doing science (and teaching)! On Mon, Feb 24, 2020 at 9:47 AM Franck Nadaud <nadaud(a)centre-cired.fr&gt; wrote: > Dear all gretl friends, greetings from Nogent-sur-Marne (just near Paris !) > > Interesting piece on a never ending debate on modelling I guess... > > We are not near a definitive answer, as I think there is none but rather a > need for sound practices anchored onto solid theoretical foundations. It > is > here that econometricians can surely contribute, as I remember a paper of > an > american philosopher who said that econometricians could surely say sound > things to physicists... > > Anyway, I recommand to all reading (or rereading) the excellent paper of > Leo > Breiman: > > http://www2.math.uu.se/~thulin/mm/breiman.pdf > > statistical modeling: the two cultures (2002). > > By the way, as usual, when new approaches and methods nurture a lot of > silly > things and errors are again and again repeated... That is, the need to > build > the sound practices to avoid the most blatant blunders and errors as well > as > develop the rights way to tackle new problems with new tools... > > Here in France we are seeing an avalanche of spurious correlations, a > reaaly > new 150 years problem we all know, from peopple with quite pompous > titles... > > On this topic I recommand also the work of the French philosopher Giuseppe > Longo whose work on randomness and algorirthms is very intersting to all of > us. His criticism of blind application of algorithms to any data set warns > of > the deluge of spurious correlations we are seeing here... > > his site an papers: > > https://www.di.ens.fr/users/longo/ > > Fact without theory, measure without theory, theory without facts, black > box > syndrome, blind application of methods without thinking about the > relevance of > the context. General to Specific or the reverse, egg or chicken... > We must stay aware and try to clarify things in the current context. > Better to > know limits and strenghts of our methods and the dark spots of our > practices. > > Well, I am getting too long. But very interesing videos ! I will share > with my > colleagues who are pûre simulation modellers... > > Cheers dear econometricians. > > All the best and sincerely. > > Franck. > > > > > > > > > > On Sun, 23 Feb 2020, Hélio Guilherme wrote: > > > >> Well, there aren't Econometricians there :D: > >> > >> https://youtu.be/uHGlCi9jOWY > > > > Ha, quite entertaining! But which side do you think the > > econometricians should be on? > > > > Allin_______________________________________________ > > Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it > > To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it > > Website: > > https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/ > > > > > ========================================= > Franck Nadaud > CIRED > UMR 8568 CNRS - EHESS, ENPC, AgroParisTech, CIRAD > 45 bis avenue de la Belle Gabrielle > 94736 Nogent-sur-Marne Cedex > TEL: 33-1-43-94-73-94 > FAX: 33-1-43-94-73-70 > France > ========================================= > > _______________________________________________ > Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it > To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it > Website: > https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/ > _______________________________________________ Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it Website: https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/