People are dating all wrong, according to data science

Joel arranged a talk at the University of Waterloo in Canada in October 2019 with a straightforward title: “Can we help people choose a better romantic partner?”

So, can Samantha Joel work with 85 of the world’s most famous scientists, combine data from 43 studies, mine hundreds of variables collected from more than 10,000, and leverage state-of-the-art machine learning models to help people be better choose a romantic partner?

Do not.

Samantha Joel told me in a Zoom interview that the first lesson in the data—and the most surprising—is “how unpredictable relationships seem to be.” Joel and her co-authors found that two people’s demographics, preferences and values ​​had little power in predicting whether two people would be happy in a relationship.

Guys, you have it. Let an AI figure out whether a group of two people can build a happy life together, and it knows as little as the rest of us.

hmm…that It does look disappointing. Can data science really fail to choose a romantic partner for us, perhaps the most important decision we will ever face in our lifetime?

incomplete. In fact, there are important lessons in Joel and her co-authors’ machine learning project, even if computers are less able to predict romantic success than many of us might have guessed.

On the one hand, while Joel and her team found that all the variables they collected had very little ability to predict couples’ happiness, they did find some variables in partners that at least slightly increased your odds of being happy with them. What’s more, the surprising difficulty of predicting romantic success has counterintuitive implications for how we should choose a romantic partner.

think about it. Many people certainly believe that many of the variables that Joel and her team studied are important in choosing a romantic partner. They compete fiercely for mates with certain traits that they assume will make them happy. If on average, as Joel and her co-authors found, many of the most competitive traits in the dating market have nothing to do with romantic happiness, it suggests that many people are dating the wrong thing.

This brings us to another age-old question that has also recently been attacked by revolutionary new data: How do people choose a romantic partner?

Over the past few years, other research groups have mined online dating sites, combing through large new data sets on the characteristics and swipes of tens of thousands of single people to determine what predicts romantic desirability. Unlike studies of romantic happiness, the results of romantic desirability studies are conclusive. While data scientists found it difficult to detect qualities that bring happiness to romantic partners, data scientists found that detecting qualities of catnip in dating scenarios was very easy.

In fact, a recent study found that not only can it be very accurate to predict whether someone will swipe at a particular person on an online dating site, but even when it will be. Someone needs to swipe the card. (People tend to take longer to find someone close to the dating acceptability threshold.)

Source link