Game of Thrones season 8: researchers create an algorithm to predict who dies



[ad_1]

Dead in Game of thrones has always been gloriously unpredictable. Just when you start to like a character (or, at least, to respect it reluctantly), they find themselves beheaded, impaled, grilled or exploded, leaving you perplexed: who is the next?

While the last season of the show is underway, a team from the Technical University of Munich (TUM) has attempted to answer this question using sophisticated database and machine learning techniques.

Their best prognosis for surviving season eight? Daenerys Targaryen, the mother of the dragons herself, with a low probability of dying of 0.9%. The character most likely to hit the medieval bucket, meanwhile, is everyone's favorite sword: Blackwater's Ser Bronn, with a 93.5% chance of dying.

Warning: Spoilers coming up for the previous Game of Thrones seasons.


Will Bronn ever have his castle? The algorithm says that the prospects are not so good.
Image: HBO

To make their predictions, the TUM team used familiar approaches to medicine and life insurance. They extracted statistical information on the life span of people, as well as biographical data likely to correspond to their death. In real life, this could include information such as whether a person smokes or how often he is exercising. But in the world of Game of thronesmore important is to know to which house belongs a particular character, whether he is married or not, and who are his allies.

With the help of wikis maintained by fans, TUM scientists have explored the lives of hundreds of characters. In addition to collecting data from the universe, such as their gender and location, they also included what we might call metadata: information such as the major or minor character of a person and the frequency with which they are cited in Wikipedias of fans.

These data revealed some fundamental truths about mortality and Game of thrones universe, like the fact that being a man is more dangerous than being a woman. (The male mortality rate is 22%, compared to 11% for women.) Some houses have a longer life than others, which reflects their ancestry in Westeros. As a Baratheon, for example, you are 5% more likely to die than the average character, while being a Lannister increases your chances of survival by 45%.

To turn these trends into predictions for individual characters, the team analyzed these data using two distinct models: the first used a fairly simple statistical approach called Bayesian inference, and the second relied on more simple techniques. sophisticated ones involving machine learning and neural networks.


Sansa Stark is another character with mediocre prospects, according to the very tight figures of TUM.
Image: HBO

According to these methods, the character deemed most likely to survive is Daenerys Targaryen. She is followed by Tyrion Lannister (2.5% chance of death), Varys (3.2%) and Samwell Tarly (3.3%). On the other side of the spectrum, Bronn is considered the most likely to die, followed by Gregor Clegane (80.3%), Sansa Stark (73.3%) and Bran Stark (57.8%). For a complete list of predictions, you should check the dedicated website where you can check the survival rate of each character.

The question is, of course, what is the quality of these predictions? It's impossible to say for sure, but it's certainly not a bad start if you feel like you're betting on Season 8.

The TUM team created a similar model in 2016, predicting that the five characters most likely to die were Tommen Baratheon (97% chance of dying), Stannis Baratheon (96%), Daenerys Targaryen (95%) , Davos Seaworth (91%) and Petyr Baelish (91%).

Looking back, we can see that most (but not all) of these predictions came true: Tommen jumped out the window at the end of Season 6, Stannis was finished by Brienne of Tarth at the end of the season 5 and Petyr was awarded the Stark Justice Award at the end of season 7. But Dany and Davos are still very much alive and fit, despite the dark predictions of the algorithm, showing the ultimate fallibility of such models.

Talk to Wired.co.ukResearcher Christian Dallago, one of the scientists involved in the project, noted that the predictions were somewhat confusing when the plot of the TV show went beyond the sources of the writer George R.R. Martin, A song of ice and fire.

"So, Daenerys currently estimates, on the whole, a 0.9% probability of dying." Three years ago, we predicted that she would be likely to die to 95%, "said Dallago. . "But since that season, George R.R. Martin has lost control and other writers are writing history. It's a little different from what was originally planned and it seems to have really affected the data. "

One of The edgeResident of Eu The experts, agreed Chaim Gartenberg, noting that the death of the series was much more unpredictable if we followed Martin's "vicious hand." Over the course of the series, some characters, like Daenerys, have become much safer, simply because they are essential to advance the plot.

Moreover, since the models created by the researchers use historical data to make their forecasts, they are necessarily blind to future political developments. For example, while being a member of the ruling clan, House Lannister, could be good for your survival rate nowOnce the different armies at play in Season 8 start walking on King's Landing, it could become a bit of a handicap.

In other words: the last movements in Game of thrones are just beginning. Do not be too sure of predictions, even if they come from an algorithm.

[ad_2]

Source link