Forgot your password?
typodupeerror

Please create an account to participate in the Slashdot moderation system

Windows

Microsoft Introduces Build Cadence Selection With Windows 10 112

Posted by Soulskill
from the putting-a-ring-on-it dept.
jones_supa writes: Microsoft has just released Windows 10 TP build 9860. Along with the new release, Microsoft is introducing an interesting cadence option for how quickly you will receive new builds. The "ring progression" goes from development, to testing, to release. By being in the slow cadence, you will get more stable builds, but they will arrive less often. By choosing the fast option, it allows you to receive the build on the same day that it is released. As a quick stats update, to date Microsoft has received over 250,000 pieces of feedback through the Windows Feedback tool, 25,381 community forum posts, and 641 suggestions in the Windows Suggestion Box.
Stats

Your Online TV Watching Can Now Be Tracked Across Devices 126

Posted by timothy
from the explodes-if-you-watch-minority-report dept.
itwbennett (1594911) writes A partnership between TV measurement company Nielsen and analytics provider Adobe, announced today, will let broadcasters see (in aggregate and anonymized) how people interact with digital video between devices — for example if you begin watching a show on Netflix on your laptop, then switch to a Roku set-top box to finish it. The information learned will help broadcasters decide what to charge advertisers, and deliver targeted ads to viewers. Broadcasters can use the new Nielsen Digital Content Ratings, as they're called, beginning early next year. Early users include ESPN, Sony Pictures Television, Turner Broadcasting and Viacom.
Microsoft

Data From Windows 10 Feedback Tool Exposes Problem Areas 139

Posted by Soulskill
from the self-awareness-kernel-keeps-uninstalling-morality-drivers dept.
jones_supa writes: Two weeks in, and already a million people have tried out Windows 10 Technical Preview, reports Microsoft, along with a nice stack of other stats and feedback. Only 36% of installations are occurring inside a virtual machine. 68% of Windows 10 Technical Preview users are launching more than seven apps per day, with somewhere around 25% of testers using Windows 10 as their daily driver (26 app launches or more per day). With the help of Windows 10's built-in feedback tool, thousands of testers have made it very clear that Microsoft's new OS still has lots of irksome bugs and misses many much-needed features. ExtremeTech has posted an interesting list of the most popular gripes received, them mostly being various GUI endurances. What has your experience been with the Technical Preview?
Privacy

The Correct Response To Photo Hack Victim-Blamers 622

Posted by samzenpus
from the who's-to-blame dept.
Bennett Haselton writes As commenters continue to blame Jennifer Lawrence and other celebrities for allowing their nude photos to be stolen, there is only one rebuttal to the victim-blaming which actually makes sense: that for the celebrities taking their nude selfies, the probable benefits of their actions outweighed the probable negatives. Most of the other rebuttals being offered, are logically incoherent, and, as such, are not likely to change the minds of the victim-blamers. Read below to see what Bennett has to say.
Math

Statisticians Uncover What Makes For a Stable Marriage 447

Posted by samzenpus
from the by-the-numbers dept.
HughPickens.com writes Randy Olson, a Computer Science grad student who works with data visualizations, writes about seven of the biggest factors that predict what makes for a long term stable marriage in America. Olson took the results of a study that polled thousands of recently married and divorced Americans and and asked them dozens of questions about their marriage (PDF): How long they were dating, how long they were engaged, etc. After running this data through a multivariate model, the authors were able to calculate the factors that best predicted whether a marriage would end in divorce. "What struck me about this study is that it basically laid out what makes for a stable marriage in the US," writes Olson. Here are some of the biggest factors:

How long you were dating: (Couples who dated 1-2 years before their engagement were 20% less likely to end up divorced than couples who dated less than a year before getting engaged. Couples who dated 3 years or more are 39% less likely to get divorced.); How much money you make: (The more money you and your partner make, the less likely you are to ultimately file for divorce. Couples who earn $125K per year are 51% less likely to divorce than couples making 0 — 25k); How often you go to church: (Couples who never go to church are 2x more likely to divorce than regular churchgoers.); Your attitude toward your partner: (Men are 1.5x more likely to end up divorced when they care more about their partner's looks, and women are 1.6x more likely to end up divorced when they care more about their partner's wealth.); How many people attended the wedding: ("Crazy enough, your wedding ceremony has a huge impact on the long-term stability of your marriage. Perhaps the biggest factor is how many people attend your wedding: Couples who elope are 12.5x more likely to end up divorced than couples who get married at a wedding with 200+ people."); How much you spent on the wedding: (The more you spend on your wedding, the more likely you'll end up divorced.); Whether you had a honeymoon: (Couples who had a honeymoon are 41% less likely to divorce than those who had no honeymoon)

Of course correlation is not causation. For example, expensive weddings may simply attract the kind of immature and narcissistic people who are less likely to sustain a successful marriage and such people might end up getting divorced even if they married cheaply. But "the particularly scary part here is that the average cost of a wedding in the U.S. is well over $30,000," says Olson, "which doesn't bode well for the future of American marriages."
Education

Despite Push From Tech Giants, AP CS Exam Counts Don't Budge Much In Most States 144

Posted by timothy
from the can't-argue-with-a-pig dept.
theodp (442580) writes "Well, the College Board has posted the 2014 AP Computer Science Test scores. So, before the press rushes out another set of Not-One-Girl-In-Wyoming-Took-an-AP-CS-Exam stories, let's point out that no Wyoming students of either gender took an AP CS exam again in 2014 (.xlsx). At the overall level, the final numbers have changed somewhat (back-of-the-Excel-envelope calculations, no warranty expressed or implied!), but tell pretty much the same story as the preliminary figures — the number of overall AP CS test takers increased, while pass rates decreased despite efforts to cherry pick students with a high likelihood of success. What is kind of surprising is how little the test numbers budged for most states — only 8 states managed to add more than 100 girls to the AP CS test taker rolls — despite the PR push by the tech giants, including Microsoft, Google, and, Facebook. Also worth noting are some big percentage decreases at the top end of the score segments (5 and 4), and still-way-too-wide gaps that exist between the score distributions of the College Board's various ethnic segments (more back of the envelope calcs). If there's a Data Scientist in the house, AP CS exam figures grabbed from the College Board's Excel 2013 and 2014 worksheets can be found here (Google Sheets) together with the (unwalkedthrough) VBA code that was used to collect it. Post your insight (and code/data fixes) in the comments!"
The Almighty Buck

Mining Kickstarter Data Reveals How To Match Crowdfunding Projects To Investors 20

Posted by timothy
from the arbitrage-opportunities dept.
KentuckyFC writes Since 2001, crowdfunding sites have raised almost $3 billion and in 2012 alone, successfully funded more than 1 million projects. But while many projects succeed, far more fail. The reasons for failure are varied and many but one of the most commonly cited is the inability to match a project with suitable investors. Now a group of researchers from Yahoo Labs and the University of Cambridge have mined data from Kickstarter to discover how investors choose projects to back. They studied over 1000 projects in the US funded by over 80,000 investors. They conclude that there are two types of backers: occasional investors who tend to back arts-related projects, probably because of some kind of social connection to the proposers; and frequent investors who have a much more stringent set of criteria. Frequent investors tend to fund projects that are well-managed, have high pledging goals, are global, grow quickly, and match their interests. The team is now working on a website that will create a list of the Twitter handles of potential investors given the URL of a Kickstarter project.
Transportation

Fuel Efficiency Numbers Overstate MPG More For Cars With Small Engines 403

Posted by Soulskill
from the emperor's-new-hybrid dept.
whoever57 writes: All official numbers for fuel economy in the EU typically overstate the miles-per-gallon figure that drivers can expect to achieve in typical driving. A recent study confirmed this once again. However, what the study also found was that MPG figures are more unrealistic for cars with smaller engines than for cars with larger engines. Actual MPG figures achieved based on typical drives for cars with small engines could be as much as 36% under the official number, while those cars with 3-liter engines would typically achieve 15% less than the official figure. These discrepancies need to be accounted for if we're going to be serious about regulating fuel efficiency. But then, we should be using gallons-per-mile instead of miles-per-gallon, too.
Advertising

Why Do Contextual Ads Fail? 249

Posted by timothy
from the pandemic-tone-deafness dept.
minstrelmike writes If we give up all our privacy on-line for contextual ads, then how come so many of them are so far off the mark? Personal data harvesting for contextual ads and content should be a beautiful thing. They do it privately and securely, and it's all automated so that no human being actually learns anything about you. And then the online world becomes customized, just for you. The real problem with this scenario is that is we're paying for contextual ads and content with our personal data, but we're not getting what we pay for. Facebook advertising is off target and almost completely irrelevant. The question is: Why? Facebook has a database of our explicitly stated interests, which many users fill out voluntarily. Facebook sees what we post about. It knows who we interact with. It counts our likes, monitors our comments and even follows us around the Web. Yet, while the degree of personal data collection is extreme, the advertising seems totally random.
Stats

Nearly 700 Genetic Factors Found To Influence Human Adult Height 68

Posted by Soulskill
from the so-much-for-genetic-height-therapy dept.
damn_registrars writes: A consortium of scientists from many different countries reviewed genome-wide association study data sets of over 250,000 individuals in a search for genetic factors that influence adult height. Looking at Single Nucleotide Polymorphisms, the researchers found 697 distinct genetic markers that can explain some 20 percent of the heritability of human adult height. Previous studies had found around 180 such markers, but the larger sample set increased the ability to detect these changes, both within genes and in non-coding regions. Genes found in this set included ones from pathways not previously connected to skeletal growth.

This study is also significant for the sample size, which allows it to address whether the data from such large sets has a tendency to converge or diverge on genetic pathways; this study particularly favors the latter, which is of great utility toward studying other polygenetic conditions in the future. The original paper is likely paywalled, however the abstract is available for free and some of the collaborators behind it have other bits available for free in the meantime.
Math

Fixing Steam's User Rating Charts 93

Posted by Soulskill
from the hmm-same-distribution-as-professional-reviews dept.
lars_doucet writes: Steam's new search page lets you sort by "user rating," but the algorithm they're using is broken. For instance, a DLC pack with a single positive review appears above a major game with a 74% score and 15,000+ ratings.

The current "user rating" ranking system seems to divide everything into big semantic buckets ("Overwhelmingly Positive", "Positive", "Mixed", etc.), stack those in order, then sort each bucket's contents by the total number of reviews per game. Given that Steam reviews skew massively positive, (about half are "very positive" or higher), this is virtually indistinguishable from a standard "most popular" chart.

Luckily, there's a known solution to this problem — use statistical sampling to account for disparate numbers of user reviews, which gives "hidden gems" with statistically significant high positive ratings, but less popularity, a fighting chance against games that are already dominating the charts.
Earth

Past Measurements May Have Missed Massive Ocean Warming 423

Posted by samzenpus
from the it's-getting-hot-in-here dept.
An anonymous reader writes "Previous estimates of global ocean warming have been significantly underestimated due to historically sparse temperature data from the Southern Ocean, new research has found. From the article: "Earth's oceans have absorbed more than 90% of the warming caused by greenhouse gases, researchers estimate, with the stored heat showing up as warmer seawater. But a new analysis suggests scientists may have underestimated the size of the heat sink in the upper ocean—which could have implications for researchers trying to understand the pace and scale of past warming."
Social Networks

Online Creeps Inspire a Dating App That Hides Women's Pictures 482

Posted by timothy
from the long-walks-on-the-beach dept.
HughPickens.com (3830033) writes "Tricia Romano reports at the Seattle Times that Susie Lee and Katrina Hess have developed Siren, a new online dating app designed to protect against men inundating women with messages that are by turns gross, hilarious, objectifying and just plain sad. A 2012 experiment by Jon Millward, a data journalist, found that women were messaged 17 times more than men; the best-looking woman received 536 messages in four months, while the best-looking guy received only 38. Lee hopes to change the nature of the messages and put women in the driver's seat. As online dating options have grown, Lee noticed that her friends' frustration did, too: With every good introduction often came a slew of lewd ones. "I just started looking (at online dating options) and very quickly realized how many things are out there and how immediately my 'creepy meter' went up," Lee says. The free iPhone app, currently launched to a select market in Seattle in August, allows women to peruse men's pictures and their answers to the "Question of the Day" ("You found a magic lamp and get three wishes. What are they?") and view their Video Challenges ("Show us a hidden gem in Seattle"). If a woman is suitably impressed by a man's answers, she can make herself visible to him. Only then can he see what she looks like. "It's a far more thoughtful — and cautious — approach than the one taken by the dating app of the moment, Tinder, which is effectively a "hot or not" game, with little information beyond a few photos, age and volunteered biographical tidbits," writes Romano. "And the implicit notion that it's a "hookup" app can be uncomfortable for some women." OK Cupid's stats as illustrated by co-founder Christian Rudder give another example of how steep the curve is, when it comes to physical attractiveness vs. messages received on online dating sites.
Censorship

Could Maroney Be Prosecuted For Her Own Hacked Pictures? 274

Posted by timothy
from the one-notch-from-thoughtcrime dept.
Contributor Bennett Haselton writes with a interesting take on the recent release of racy celebrity photos: "Lawyers for Olympic gymnast McKayla Maroney succeeded in getting porn sites to take down her stolen nude photos, on the grounds that she was under 18 in the pictures, which meant they constituted child pornography. If true, that means that under current laws, Maroney could in theory be prosecuted for taking the original pictures. Maybe the laws should be changed?" Read on for the rest.
United States

Laying the Groundwork For Data-Driven Science 55

Posted by samzenpus
from the collecting-the-numbers dept.
aarondubrow writes The ability to collect and analyze massive amounts of data is transforming science, industry and everyday life. But what we've seen so far is likely just the tip of the iceberg. As part of an effort to improve the nation's capacity in data science, NSF today announced $31 million in new funding to support 17 innovative projects under the Data Infrastructure Building Blocks (DIBBs) program, including data infrastructure for education, ecology and geophysics. "Each project tests a critical component in a future data ecosystem in conjunction with a research community of users," said said Irene Qualters, division director for Advanced Cyberinfrastructure at NSF. "This assures that solutions will be applied and use-inspired."
Transportation

Which Cars Get the Most Traffic Tickets? 261

Posted by Soulskill
from the your-'84-buick-lesabre-is-probably-not-getting-you-into-trouble dept.
An anonymous reader writes: Have you ever been pulled over for a traffic stop and wondered if your sporty car was what caught the officer's attention? Ever had an officer pass up your clunker to snag a flashier vehicle? Well, there's now some data showing which vehicles accumulate the most tickets. According to a study by Insurance.com, drivers of the Subaru WRX, Pontiac GTO, and Scion FR-S get a higher percentage of tickets than drivers of any other cars. At the bottom of the list, we see vehicles such as the Ford Ecosport, the Land Rover LR4, and Chevy Sportvan. They have a widget that will let you see data on your own make/model, if you're curious.
The Internet

Analyzing Silk Road 2.0 68

Posted by Soulskill
from the welcome-to-narcoanalytics dept.
An anonymous reader writes: After a recent article about breaking the CAPTCHA on the latest incarnation of Silk Road (the darknet-enabled drug market place), Darryl Lau decided to investigate exactly what narcotics people were buying and selling online. He found roughly 13,000 separate listings. Some sellers identify the country they're in, and the top six are the U.S., Australia, England, Germany, and the Netherlands, and Canada. The site also has a bunch of product reviews. If you assume that each review comes from a sale, and multiply that by the listed prices, reviewed items alone represent $20 million worth of business. Lau also has some interesting charts, graphs, and assorted stats. MDMA is the most listed and reviewed drug, and sellers are offering it in quantities of up to a kilogram at a time. The average price for the top 1000 items is $236. Prescription drugs represent a huge portion of the total listings, though no individual prescription drugs have high volume on their own.
Math

Statistician Creates Mathematical Model To Predict the Future of Game of Thrones 127

Posted by samzenpus
from the math-is-coming dept.
KentuckyFC writes One way of predicting the future is to study data about events in the past and build a statistical model that generates the same pattern of data. Statisticians can then use the model to generate data about the future. Now one statistician has taken this art to new heights by predicting the content of the soon-to-be published novels in the Song of Ice and Fire series by George R R Martin. The existing five novels are the basis of the hit TV series Game of Thrones. Each chapter in the existing books is told from the point of view of one of the characters. So far, 24 characters have starred in this way. The statistical approach uses the distribution of characters in chapters in the first five books to predict the distribution in the forthcoming novels. The results suggest that several characters will not appear at all and also throw light on whether one important character is dead or not, following an ambiguous story line in the existing novels. However, the model also serves to highlight the shortcomings of purely statistical approaches. For example, it does not "know" that characters who have already been killed off are unlikely to appear in future chapters. Neither does it allow for new characters that might appear. Nevertheless, this statistical approach to literature could introduce the process of mathematical modelling to more people than any textbook.
Stats

Ask Slashdot: Is Reporting Still Relevant? 179

Posted by timothy
from the that-hansel-he's-so-hot-right-now-and-does-reports dept.
New submitter MrWHO (68268) writes A while ago we switched for monitoring our systems to the ELK (ElasticSearch, LogStash and Kibana) stack. Our management wanted to keep the reports they got — and possibly never read — flowing in at the beginning of every week with statistics like sites traffic, servers downtime, security alerts and the works. As we migrated some of our clients to the same stack they kept all asking for the same thing: reporting. There was no way for us to create and schedule reports from ElasticSearch — searches for ElasticSearch and Jasper Reports returned nothing apart from people asking how to do it — so we created our own Jasper Reports plugin to create reports from ElasticSearch data, which we released on GitHub a while ago, and we promptly moved along.

None of our clients were easily convinced that a dashboard — Kibana — was a substitute for mail delivered PDFs, even if all the information was there, with custom created panels and selectable date ranges. On the other hand, on the ElasticSearch mailing list when questions were asked about "how do I do reports?" the answer was, and I sum it up here, "Why would you want reports when you have a dashboard?" Are reports still relevant — the PDF, templated, straight in to your mail kind — or the subset of my clients — we operate mainly in Italy — is a skewed sample of what's the actual reality of access to summary data? Are dashboards — management targeted ones — the current accepted solution or — in your experience — reports are still a hot item for management?
Math

New Study Projects World Population of 11B by 2100 326

Posted by timothy
from the ok-but-only-one-of-me dept.
vinces99 (2792707) writes Using modern statistical tools, a new study led by the University of Washington and the United Nations finds that world population is likely to keep growing throughout the 21st century. The number of people on Earth is likely to reach 11 billion by 2100, the study concludes, about 2 billion higher than widely cited previous estimates. The paper published online Sept. 18 in the journal Science includes the most up-to-date numbers for future world population, and describes a new method for creating such estimates. "The consensus over the past 20 years or so was that world population, which is currently around 7 billion, would go up to 9 billion and level off or probably decline," said corresponding author Adrian Raftery, a UW professor of statistics and of sociology. ... The paper explains the most recent United Nations population data released in July. This is the first U.N. population report to use modern statistics, known as Bayesian statistics, that combines all available information to generate better predictions.

Most of the anticipated growth is in Africa, where population is projected to quadruple from around 1 billion today to 4 billion by the end of the century. The main reason is that birth rates in sub-Saharan Africa have not been going down as fast as had been expected. There is an 80 percent chance that the population in Africa at the end of the century will be between 3.5 billion and 5.1 billion people.

This is a good time to punt work.

Working...