Why It’s So Freaking Hard To Make A Good COVID-19 Model | FiveThirtyEight
▻https://fivethirtyeight.com/features/why-its-so-freaking-hard-to-make-a-good-covid-19-model
Why It’s So Freaking Hard To Make A Good COVID-19 Model | FiveThirtyEight
▻https://fivethirtyeight.com/features/why-its-so-freaking-hard-to-make-a-good-covid-19-model
Will Women In Low-Wage Jobs Get Their #MeToo Moment? | FiveThirtyEight
▻https://fivethirtyeight.com/features/the-metoo-moment-hasnt-reached-women-in-low-wage-jobs-will-it
▻https://espnfivethirtyeight.files.wordpress.com/2017/12/harassment-16x9.jpg?w=1600
Merci @touti, sans ton up, je l’aurais raté !
AlphaZero : l’IA de Google DeepMind devient imbattable aux échecs
▻http://www.futura-sciences.com/tech/actualites/technologie-alphazero-ia-google-deepmind-devient-imbattable-echecs-
Shut Up About Harvard | FiveThirtyEight
▻https://fivethirtyeight.com/features/shut-up-about-harvard
“Ninety-five percent of the newsroom probably went to private institutions, they went to four-year institutions, and they went to elite institutions,” said Jeff Selingo, a longtime higher-education journalist who has a new book focused on giving advice to a broader group of students. “It is exactly the opposite of the experience for the bulk of American students.”
(...)
That myopia has real consequences for education policy. (...) The media’s focus on elite schools draws attention away from state cuts to higher-education funding, for example.
Five ThirtyHeight
Gun Deaths In America
▻https://fivethirtyeight.com/features/gun-deaths
▻https://static1.squarespace.com/static/53407e27e4b0f0bdc6e38fb9/58d846d02994ca9ba7380543/58d846d159cc68feaa5325d5/1490568916171/Screen+Shot+2016-09-18+at+11.15.01+PM.png?format=1500w
“Methodology
The data in this interactive graphic comes primarily from the Centers for Disease Control and Prevention’s Multiple Cause of Death database, which is derived from death certificates from all 50 states and the District of Columbia and is widely considered the most comprehensive estimate of firearm deaths. In keeping with the CDC’s practice, deaths of non-U.S. residents that take place in the U.S. (about 50 per year) are excluded. All figures are averages from the years 2012 to 2014, except for police shootings of civilians, which are from 2014.
The “homicides” category includes deaths by both assault and legal intervention (primarily shootings by police officers). “Young men” are those ages 15 to 34; “women” are ages 15 and older. Because the CDC’s estimates of police shootings are unreliable, we used estimates from non-governmental sources. Our figure is for 2014, the first year for which such estimates are generally available. (For more on the data we used, see Carl Bialik’s story on police shootings.)
For shootings of police officers, we used the FBI’s count of law enforcement officers “feloniously killed” by firearms in the line of duty. This figure excludes accidental shootings. The FBI counts all killings of federal, state and local law enforcement officers who meet certain criteria, including that they were sworn officers who ordinarily carried a badge and a gun.
For mass shootings, we used Mother Jones’s database of public mass shootings. For 2012 and earlier, Mother Jones includes only incidents in which at least four people (excluding the shooter) were killed; beginning in 2013, Mother Jones lowered the threshold to three fatalities. In order to use a consistent definition, we excluded the one incident in 2013-14 in which exactly three people were killed.
For terrorism gun deaths, we used the University of Maryland’s Global Terrorism Database. Our count of fatalities excludes perpetrators killed during their attacks. There was one incident, the 2012 attack on a Sikh temple in Oak Creek, Wisconsin, that qualified as both an act of terrorism and a mass shooting. Seven law-enforcement officers were killed in incidents that the terrorism database classifies as acts of terrorism.
Population totals (used to calculate death rates per 100,000 people) are based on 2012-14 American Community Survey microdata from the University of Minnesota’s IPUMS project. As a result, death rates will not perfectly match official figures from the CDC, which are based on a different set of numbers from the Census Bureau. Racial and ethnic categories are mutually exclusive: All people who were designated as Hispanic in the CDC data are coded as “Hispanic” in ours; all other racial categories are non-Hispanic. “Native American” includes American Indians and Alaska Natives.
Data and code for this project are available on our GitHub page.”
Data On Drug Use Is Disappearing Just When We Need It Most
@fil
▻https://fivethirtyeight.com/features/data-on-drug-use-is-disappearing-just-when-we-need-it-most
▻https://espnfivethirtyeight.files.wordpress.com/2017/06/casteel-drug-data-0620-1.png?quality=90&strip=i
“It’s no secret that heroin has become an epidemic in the United States. Heroin overdose deaths have risen more than sixfold in less than a decade and a half.1 Yet according to one of the most widely cited sources of data on drug use, the number of Americans using heroin has risen far more slowly, roughly doubling during the same time period.2
Most major researchers believe that source, the National Survey on Drug Use and Health, vastly understates the increase in heroin use. But many rely on the survey anyway for a simple reason: It’s the best data they have. Several other sources that researchers once relied on are no longer being updated or have become more difficult to access. The lack of data means researchers, policymakers and public health workers are facing the worst U.S. drug epidemic in a generation without essential information about the nature of the problem or its scale.
“We’re simply flying blind when it comes to data collection, and it’s costing lives,” said John Carnevale, a drug policy expert who served at the federal Office of National Drug Control Policy under both Republican and Democratic administrations. There is anecdotal evidence of how patterns of drug use are changing, Carnevale said, and special studies conducted in various localities are identifying populations of drug users. “But the national data sets we have in place now really don’t give us the answers that we need,” he said.”
Children of the Opioid Epidemic Are Flooding Foster Homes. America Is Turning a Blind Eye.
Long before the social workers showed up in his living room this March, Matt McLaughlin, a 16-year-old with diabetes, had taken to a wearying evening routine: trying to scrounge up enough spare change for food while his mom, Kelly, went to a neighbor’s house to use heroin. On a good night, the bookish high school junior would walk through his neighborhood in Andover, Ohio, a Rust Belt town surrounded by fields and trailer parks, to pick up frozen pizza from the Family Dollar. On a bad night, he’d play video games to distract himself from his grumbling stomach and dipping blood sugar, and wait for Kelly to return with glazed eyes.
Dissecting Trump’s Most Rabid Online Following, by Trevor Martin | FiveThirtyEight
▻https://fivethirtyeight.com/features/dissecting-trumps-most-rabid-online-following
▻https://espnfivethirtyeight.files.wordpress.com/2017/03/thedonaldtrumpsubredditalgebra-16x9.jpg?w=2667
Comparing subreddits, with Latent Semantic Analysis in R
▻http://blog.revolutionanalytics.com/2017/03/comparing-subreddits.html
The article looks at various popular and notorious subreddits and finds those that are most similar to the main subreddit devoted to Donald Trump and also to the main other contenders in the 2016 campaign for president, Hillary Clinton and Bernie Sanders.
#machine_learning #LSA #gensim #R #howto #trolls #reddit #text-mining
A Computer Just Clobbered Four Pros At Poker | FiveThirtyEight
▻https://fivethirtyeight.com/features/a-computer-just-clobbered-four-pros-at-poker
▻https://espnfivethirtyeight.files.wordpress.com/2017/01/roeder-poker-update-1.png?quality=90&strip=all&
About three weeks ago, I was in a Pittsburgh casino for the beginning of a 20-day man-versus-machine poker battle. Four top human pros were beginning to take on a state-of-the-art artificial intelligence program running on a brand new supercomputer in a game called heads-up no-limit Texas Hold ’em. The humans’ spirits were high as they played during the day and dissected the bot’s strategy over short ribs and glasses of wine late into the evening.
On Monday evening, however, the match ended and the human pros were in the hole about $1.8 million. For some context, the players (four men and the machine, named Libratus) began each of the 120,000 hands with $20,000 in play money, and posted blinds of $50 and $100.
...
Tuomas Sandholm, a Carnegie Mellon computer scientist who created the program with his Ph.D. student Noam Brown, was giddy last week on the match’s livestream, at one point cheering for his bot as it turned over a full house versus human pro Jason Les’s flush in a huge pot, and proudly comparing Libratus’s triumph to Deep Blue’s monumental win over Garry Kasparov in chess.
And, indeed, some robot can now etch heads-up no-limit Texas Hold ‘em (2017) alongside checkers (1995), chess (1997), Othello (1997), Scrabble (c. 2006), limit Hold ‘em (2008), Jeopardy! (2011) and Go (2016) into the marble cenotaph of human-dominated intellectual pursuits.
Brown told me that he was keen to tackle other versions of poker with his A.I. algorithms. What happens when a bot like this sits down at a table with many other players, rather than just a one-on-one foe, for example? Sandholm, on the other hand, is quick to say that this isn’t really about poker at all. “The AI’s algorithms are not for poker: they are game independent,” his daily email updates read. The other “games” the algorithms may be applied to in the future: “negotiation, cybersecurity, military setting, auctions, finance, strategic pricing, as well as steering evolution and biological adaptation.”
Pourquoi les commentateurs commentent-ils sur le Web ?
▻http://lemonde.fr/big-browser/article/2016/11/30/troller-debattre-ou-se-repeter-pourquoi-commenter-un-article-en-ligne_504099
A travers un questionnaire rempli par 8 500 personnes, FiveThirtyEight dresse le portrait de ses commentateurs et de leurs motivations, un travail qui ne saurait être représentatif de l’ensemble d’Internet, mais dans lequel on trouve des pistes familières :
19 % d’entre eux interviennent pour « corriger une erreur ».
18 % pour « ajouter quelque chose à la conversation ».
10 % pour « donner leur perspective personnelle ».
Certains articles déclenchent plus de commentaires que d’autres, et il semble presque impossible, en tout cas empiriquement, de savoir à l’avance si un article va inspirer les commentateurs. FiveThirtyEight a donc demandé aux lecteurs quelles étaient les circonstances dans lesquelles ils avaient le plus de chance de commenter.
=> ▻https://fivethirtyeight.com/features/we-asked-8500-internet-commenters-why-they-do-what-they-do
[Sont-ce les mêmes qui répondent aux questionnaires sur Internet ?]
Dans 55 % des cas, c’est quand ils « savent quelque chose sur le sujet qui n’était pas dans l’article », et 41 % lorsqu’ils « s’identifient au sujet abordé ». Sans surprise, les lecteurs ont donc tendance à s’engager dans la « conversation » quand le sujet les concerne ou qu’ils ont l’impression d’avoir quelque chose de nouveau à apporter.
These Are The Phrases That Sanders And Clinton Repeat Most ou “le fond et la forme” | FiveThirtyEight
▻https://fivethirtyeight.com/features/these-are-the-phrases-that-sanders-and-clinton-repeat-most
Thousands Of Ted Cruz Supporters Play A Game That Might Win Iowa
▻https://fivethirtyeight.com/features/thousands-of-ted-cruz-supporters-play-a-game-that-might-win-iowa
▻https://espnfivethirtyeight.files.wordpress.com/2015/12/ap_242409902451.jpg?w=1200
The Cruz mobile app, which game-ifies the campaign process — users win points for sharing their contact lists or for posting messages of support to Facebook and Twitter — has close to 30,000 downloads, according to Chris Wilson, head of analytics for the campaign.
In a rural state like Iowa, some supporters say they’ve developed a sense of community from the app. Linda Stickle lives in Stone City, an eastern Iowa town that Grant Wood of American Gothic fame once painted in all its undulating stillness. “It’s a suburb of Anamosa,” Stickle said, adding wryly, “Anamosa has 5,000 people.”
Stickle, 66, whose husband farms, has a number of what she calls “rural contacts” — she spent a summer delivering bulls via truck to farmers all over the state. She has acted as a Cruz proxy at gatherings in her county. “I have met so many wonderful people and I have made relationships, just doing this,” Stickle said.
Her friend Kay Quirk, 61, of Buena Vista County, in the northwestern part of the state, agreed. “My husband says it’s a good thing we don’t live closer to you because then he would never see me!” she said to Stickle.
Quirk holds the No. 1 slot on the Cruz Crew Iowa leaderboard with nearly 172,000 points (Stickle is No. 3 in the state with over 30,000) and says she devotes a couple of hours a day to posting to Twitter and Facebook. “Ted Cruz called me the ‘Twitter warrior,’ ” she said.
10 Years After Katrina - The New York Times
▻http://www.nytimes.com/interactive/2015/08/26/us/ten-years-after-katrina.html?_r=0
NEW ORLEANS — It is a wonder that any of it is here at all: The scattered faithful gathering into Beulah Land Baptist Church on a Sunday morning in the Lower Ninth Ward. The men on stoops in Mid-City swapping gossip in the August dusk. The brass band in Tremé, the lawyers in Lakeview, the new homeowners in Pontchartrain Park.
On Aug. 29, 2005, it all seemed lost. Four-fifths of the city lay submerged as residents frantically signaled for help from their rooftops and thousands were stranded at the Superdome, a congregation of the desperate and poor. From the moment the storm surge of Hurricane Katrina dismantled a fatally defective levee system, New Orleans became a global symbol of American dysfunction and government negligence. At every level and in every duty, from engineering to social policy to basic logistics, there were revelations of malfunction and failure before, during, and after Katrina.
Ten years later, it is not exactly right to say that New Orleans is back. The city did not return, not as it was.
It is, first of all, without the more than 1,400 people who died here, and the thousands who are now making their lives someplace else. As of 2013, there were nearly 100,000 fewer black residents than in 2000, their absences falling equally across income levels. The white population decreased by about 11,000, but it is wealthier.
The city that exists in 2015 has been altered, by both a decade of institutional re-engineering and the artless rearrangement that occurs when people are left to fend for themselves.
Empowered by billions of federal dollars and the big ideas of eager policy planners, the school system underwent an extensive overhaul; the old Art Deco Charity Hospital was supplanted by a state-of-the-art medical complex; and big public housing projects, at once beloved and notorious, were razed and replaced by mixed-income communities with housing vouchers.
#Nouvelle_Orleans #Katrina #gentrification #remplacement #data #interaction #cartes
We Still Don’t Know How Many People Died Because Of Katrina | FiveThirtyEight
▻https://fivethirtyeight.com/features/we-still-dont-know-how-many-people-died-because-of-katrina
▻https://espnfivethirtyeight.files.wordpress.com/2015/08/bialikkatrinafeatured.jpg?quality=100&strip=all
We Still Don’t Know How Many People Died Because Of Katrina
By Carl Bialik
Ten years after Hurricane Katrina struck the Gulf Coast and the New Orleans levees failed, we still don’t know how many people died in the storm and its aftermath.
The uncertainty about the death toll is evident in the variety of numbers being reported by the media. A local news station in Georgia: 1,200. AccuWeather: 1,800. Insurance Journal: more than 1,800. The New Orleans Times-Picayune: 1,833. A local news station in western Michigan: 1,836.
There is still no memorial listing the names of Katrina victims, still no way to know how many remain uncounted or unidentified, and still no agreement on how to count victims if a storm of Katrina’s impact hits the U.S. again. Ten years on, we’re still in the dark.