Mathematician cracked dating and found a girl dreams

35-year-old math Chris McKinley (Chris McKinlay) from the University of California in Los Angeles, it took only 88 days to implement the data-mining profiles on a dating site and find the girl who is perfect in nature. This helped him access to supercomputers, University of California, writes Wired.

In June 2012, McKinley was working on his PhD in processing large amounts of data and methods of parallel computing. At some point, he decided to escape and join a dating site OkCupid. It should be noted that this site was founded by Harvard students and uses cunning calculation system compatibility between potential partners. Each of them is responsible for hundreds of questions about his character, habits and preferences. A variety of questions: “Do you like cats?” (What answers you accept from others), “How often do you use Facebook?”, “How long are planning to meet?” etc. Then you can see how this or that person is right for you. The system shows the probability of a successful relationship, friendship and enmity. The logic is that the more similar partners, the better they will be with each other.

The trick is to look at how she replied to a specific question, you can only after you have answered it. The maximum amount for each participant is limited to 350 responses, while total bank contains thousands of questions. In other words, some answers girls will forever remain a mystery to you, and “compatibility rating” may be slightly inaccurate.

McKinley found only about 100 girls who have compatibility rate exceeded 90%, but even with them goodbye somehow did not work out (Chris enough for six visits). And then the guy came up with a way to analyze their own rules.

Chris created the fake profiles 12, answering questions randomly. After that, he was able to get access to the full profiles of girls with all the answers. For three weeks the date of its Mining System collected 6 million answers from a test sample of 20 thousand questionnaires.

After filtration, the remaining 5,000 women, who live in Los Angeles and San Francisco and visited the site in the last month. Chris wrote a Python script to sort the girls with different characteristics into seven clusters, such as “Caring”, “Believers”, “tattoo”, etc.

Two clusters he found particularly interesting. The first was a girl about 25 years hipovatogo style, musicians and actresses. The second cluster includes the older girl with creative professions in the literature and design. So Chris created two of its profile to focus on each of these clusters - and responded to 500 most popular questions in each of the groups. Here is Chris answered the questions honestly, although slightly optimized answers depending on the cluster …

25 January 2014

95 percent of U.S. ATMs run on Windows XP
Order poster and get a gift car LADA GRANTA

• Rating Runet. Results of ranking web studios »»»
Project Rating Runet published a list of the leading web studios Russia in 2009.
• In Russia, after two years of development launched single portal on the state budget $ 1 billion »»»
Single portal budget system, where you can find all the information about income and expenditure in the public sector, has launched a test operation.
• "Kaspersky Lab" has begun testing its own operating system »»»
Integrators - partners are beta testing the OS protected industrial "Kaspersky Lab", a project known as "11.
• Twitter suggested MOE notify Russians about PE tweets »»»
Twitter will contact the Russian Consulate General in San Francisco with a proposal to create in our country a special twitter account for emergency notification.
• Objective rating of the content management system "CMS Magazine" »»»
Project “CMS Magazine” presents a completely new tool for analyzing the content management system.