Les bookmarks de Clem
2353 shaares
Via Mathieu.
À noter, yen a plein d'autres ici: http://www.chrd.lyon.fr/chrd/sections/fr/pages_fantomes/collections/affiches/
À noter, yen a plein d'autres ici: http://www.chrd.lyon.fr/chrd/sections/fr/pages_fantomes/collections/affiches/
Dans la série travail déclassifié de Turing pendant la seconde guerre mondiale.
Pas mal de liens intéressant sur ce thème. À lire...
Andrew Gelman sur le travail déclassifié de Turing concernant les probabilités. L'article de Turing a l'air ultra intéressant!
Ça chauffe sur twitter...
Il vaut mieux chier accroupi que assis, c'est meilleur pour la santé.
La liste des journaux en écologie qui acceptent et refusent le LaTeX.
Très intéressant: les structures organisationnelles prennent le pouvoir dans toute organisation. Ceux qui bossent à atteindre les objectifs de l'organisation seront toujours dirigés par ceux qui vont prendre en charge l'organisation.
Un bon scientifique doit savoir dire "je ne sais pas"...
Intéressant. Le statisticien est interviewé sur le plantage des sondages pré-élections en GB, et on lui pose une question de sociologie (les sondages influent-ils sur le vote des gens), il ne sait pas répondre, s'embrouille, patauge et essaie de revenir dans son domaine de compétence. Après coup, il réfléchit et écrit cet article. Il note que dans ce type d'interview, il y a un patron général: "The common factor is this: the interviewer wants to turn the discussion either to:
* the effect the numbers have on people, or
* why people affect the numbers.
I cannot criticise them - it is entirely understandable that they are interested in the human story around the stats, and it probably reflects what the audience would ask. But all this is generally outside the expertise of the statistician. It doesn’t seem fair: astronomers don't expect to be asked about the effect their discoveries might have on people. But we statisticians clearly have to be ready."
Deux stratégies alors: (i) to have done your homework and spent some time examining the human context of the numbers, and at least be ready to summarise what social scientists have said about people’s behaviour. (...) (ii) to not only be ready for the question that you are not qualified to answer, but to positively welcome it. It gives a chance to explain that (to parody an old cliché) science means not having to say you know. It is OK not to have opinions about things until you have studied the evidence, and even then the conclusions may not be clear.
Edit: le statisticien en question est David Spiegelhalter.
* the effect the numbers have on people, or
* why people affect the numbers.
I cannot criticise them - it is entirely understandable that they are interested in the human story around the stats, and it probably reflects what the audience would ask. But all this is generally outside the expertise of the statistician. It doesn’t seem fair: astronomers don't expect to be asked about the effect their discoveries might have on people. But we statisticians clearly have to be ready."
Deux stratégies alors: (i) to have done your homework and spent some time examining the human context of the numbers, and at least be ready to summarise what social scientists have said about people’s behaviour. (...) (ii) to not only be ready for the question that you are not qualified to answer, but to positively welcome it. It gives a chance to explain that (to parody an old cliché) science means not having to say you know. It is OK not to have opinions about things until you have studied the evidence, and even then the conclusions may not be clear.
Edit: le statisticien en question est David Spiegelhalter.
Ah bon? c'est faux grammaticalement de dire au final?
Bon, va falloir que je change mes habitudes alors (pis celle-là elle est bien ancrée!).
Bon, va falloir que je change mes habitudes alors (pis celle-là elle est bien ancrée!).
Une étude monstrueuse de réplication de résultats en psychologie: sur 100 études reproduites, 39 étaient reproductibles d'après des critères définis à priori (et 24 de plus si l'on oublie ces critères de significativité etc., et que l'on considère le message de chaque étude). Pas encore peer-reviewed, mais ça sortira dans science.
Entre 40 et 60% de reproductibilité, ça confirme un point déjà noté précédement: la moitié de la littérature scientifique est fausse...
Ça fait mal quand même...
À suivre
Entre 40 et 60% de reproductibilité, ça confirme un point déjà noté précédement: la moitié de la littérature scientifique est fausse...
Ça fait mal quand même...
À suivre
Encore un xkcd de génie
Un générateur de nombre aléatoire en ligne. De nombres vraiment aléatoires, basés sur un générateur physique, pas des nombres pseudo-aléatoires.
Et chose marrante, ya même un package R qui peut se connecter à ce service pour générer des vrais nombres aléatoires (package random).
Et chose marrante, ya même un package R qui peut se connecter à ce service pour générer des vrais nombres aléatoires (package random).
Tiens? un article qui a l'air super intéressant.
À lire absolument
À lire absolument
M'a l'air intéressant. À lire
Ben putain, ça fait mal. En vrac:
"Preventive medicine displays all 3 elements of arrogance. First, it is aggressively assertive, pursuing symptomless individuals and telling them what they must do to remain healthy. (...) Second, preventive medicine is presumptuous, confident that the interventions it espouses will, on average, do more good than harm to those who accept and adhere to them. Finally, preventive medicine is overbearing, attacking those who question the value of its recommendations."
"Without evidence from positive randomized trials (and, better still, systematic reviews of randomized trials) we cannot justify soliciting the well to accept any personal health intervention. There are simply too many examples of the disastrous inadequacy of lesser evidence as a basis for individual interventions among the well: supplemental oxygen for healthy premies (causing retrolental fibroplasia), healthy babies sleeping face down (causing SIDS), thymic irradiation in healthy children, and the list goes on."
On rajoute à ça la prescription d'hormones aux femmes ménopausées, recommandées jusqu'à très récemment, et dont une étude a pu montrer que le traitement était calamiteux!
"I place the blame directly on the medical “experts” who, to gain private profit (from their industry affiliations), to satisfy a narcissistic need for public acclaim or in a misguided attempt to do good, advocate “preventive” manoeuvres that have never been validated in rigorous randomized trials."
"Preventive medicine displays all 3 elements of arrogance. First, it is aggressively assertive, pursuing symptomless individuals and telling them what they must do to remain healthy. (...) Second, preventive medicine is presumptuous, confident that the interventions it espouses will, on average, do more good than harm to those who accept and adhere to them. Finally, preventive medicine is overbearing, attacking those who question the value of its recommendations."
"Without evidence from positive randomized trials (and, better still, systematic reviews of randomized trials) we cannot justify soliciting the well to accept any personal health intervention. There are simply too many examples of the disastrous inadequacy of lesser evidence as a basis for individual interventions among the well: supplemental oxygen for healthy premies (causing retrolental fibroplasia), healthy babies sleeping face down (causing SIDS), thymic irradiation in healthy children, and the list goes on."
On rajoute à ça la prescription d'hormones aux femmes ménopausées, recommandées jusqu'à très récemment, et dont une étude a pu montrer que le traitement était calamiteux!
"I place the blame directly on the medical “experts” who, to gain private profit (from their industry affiliations), to satisfy a narcissistic need for public acclaim or in a misguided attempt to do good, advocate “preventive” manoeuvres that have never been validated in rigorous randomized trials."
Excellent article. Petit passage également valable pour certains utilisateurs de logiciels: "To answer a single email from a student–either in the form of a long list of questions or just an open-ended plea for help–takes a lot of time. We may respond to the first few emails we get, but as they keep pouring in, we tend to burn out. And the more popular this becomes as a pedagogical tool, the more emails students will be sending to scientists and writers. And that makes people burn out even faster. It doesn’t seem fair to the students for their grade to depend on whether they get a reply from their email. *Even the most polite email may land in the inbox of someone who decided long ago never to respond to such requests*."
(l'emphase est de moi: c'est mon cas)
(l'emphase est de moi: c'est mon cas)
Un joli comparatif comparant R et Python. Avantages et inconvénients de chacun. Sur le long-terme, python pourrait bien passer premier. Pour le moment, on en est encore loin...
Un tutoriel super intéressant sur chroot (via sebsauvage)