Predicting the Unexpected: A Bayesian Model to Predict the Winner of Big Brother 18

Summer is here, which means it’s once again time for everyone’s favorite Julie Chen-hosted reality show, Big Brother.

Okay, maybe it’s not everyone’s favorite reality show. Maybe it’s not even anyone’s favorite reality show. But watching Big Brother every summer has become something of a tradition in my house. My wife and I have watched every season together for the last 13 years. It’s sort of a TV guilty pleasure. You know, like anything on Bravo or CNN.

If you’ve never seen Big Brother, which currently airs every Sunday, Wednesday, and Thursday on CBS, here’s a brief description, courtesy of Wikipedia:

Big Brother is a television reality game show based on an originally Dutch TV series of the same name created by producer John de Mol in 1997. The series follows a group of contestants, known as House Guests, who are living together in a custom-built home under constant surveillance. The House Guests are completely isolated from the outside world, and can have no communication with those not in the house. The contestants are competing for a $500,000 grand prize, with weekly competitions and evictions determining who will win the show. The series takes its name from the character in George Orwell’s novel Nineteen Eighty-Four (1984).

So what’s the appeal, you ask? Despite occasional goofiness and borderline absurdity (just Google “Zingbot”), there’s an entertaining – and actually quite gripping, at times – psychological component to the show. In order to avoid eviction and win the $500,000 prize on finale night, House Guests must compete, scheme, plot, form alliances, make deals, and sometimes even stab former friends and allies in the back. Indeed, if you watch the show, a phrase you’ll likely hear over and over again is “thrown under the bus,” as in “so-and-so was thrown under the bus by her entire alliance.” So, I suppose you can think of the show as the intersection of chess, poker, and MTV’s Real World. Maybe with even a little Jerry Springer thrown in for good measure. Because, of course, each season is bound to have its fair share of fights and “showmances.”

Now, I don’t plan to inundate you, dear reader, with a steady stream of blog posts over the next 12-13 weeks detailing everything that’s happening inside the Big Brother House. So have no fear. For one thing, I’m too cheap to subscribe to the live feeds. Secondly, there are far better places to go for that level of fandom and nerdiness.

But for the unique segment of the general population that is a fan of both Julie Chen and Data Science, I can provide another service, one that is no less nerdy. Yes, I’ve gone ahead and put together a statistical model that will attempt to predict the eventual winner of the current season, Big Brother 18.

The model uses Bayesian Inference and data from the past 15 seasons (Seasons 3-17) to look for the current House Guest whose game play most closely mirrors that of previous show winners. Quite simply, the House Guest whose performance most closely resembles that of previous winners is assigned the highest probability of winning.

I’ll update the model several times a week based on the following five events (each of which occurs on a weekly basis inside the Big Brother House):

  1. Who won the Head-of-Household (HoH) Competition for the week?
  2. Who was nominated for eviction during the initial Nomination Ceremony?
  3. Who won the Power of Veto (PoV) Competition?
  4. Who was nominated for eviction following the Veto Competition?
  5. Who was evicted from the house on Eviction Night?

Importantly, the producers of the show try to incorporate unexpected twists and turns each season. Hence, the tagline for the show: “Expect the Unexpected.”

For instance, this season there is the BB Roadkill twist, which allows for the nomination of a third House Guest each week for the first several weeks of the season. I’m not including the outcome of the BB Roadkill challenge in my model because there’s no prior data on how winning this challenge affects a House Guest’s chances of winning the grand prize. But I will obviously include the fact that there will be three nominees for eviction during each of the first few weeks.

It’s anyone’s guess as to whether this model will go on to predict the eventual winner. The unique twists and challenges of each season might make it ultimately impossible to use data from prior seasons to make predictions about the current season. But it will be fun and interesting to see, regardless. Even if the model fails spectacularly.

So, check back here regularly if you’re a fan of the show and if you want to know who currently has the best shot at walking away with $500,000 at the end of the summer.

And if you have any comments or questions, feel free to leave them below.

 

BIG BROTHER 18 – MODEL PREDICTIONS

Which House Guest has the best chance of winning Big Brother 18?

 

Brian Kurilla is a psychological scientist with a Ph.D. in cognitive psychology. You can follow Brian on Twitter @briankurilla 

Leave a Reply

Your email address will not be published. Required fields are marked *