During the long, long winters in between Game Of Thrones seasons, fans love to speculate who'll die next -- and who should die from their sheer stupidity. Now, one computer programmer claims he has found an algorithm to successfully augur the likelihood of which main characters will soon be meeting their Lord of Light.
Taylor Larkin, a "data science evangelist" working for DataRobot, employed a machine learning technique called gradient boosting (which can use wide decision trees to map probability) to calculate the chances of death of our most beloved GOT characters and Bran. The algorithm is based on information on a character's house, culture, title, status, gender, and family mortality rate -- which must be why the all the remaining Starks just got a "yikes" out of 10.
Some of the biggest losers according to DataRobot are Daenerys Targaryen (who has a 83.77 percent chance of death) and Jaime Lannister (72.91 percent, one point for each pound of his massive golden hand). To get the necessary info, the machine scraped the data of over 2,000 characters logged in the biggest A Song Of Ice And Fire wiki. So its findings are based mostly on the books instead of the TV show, which explains why the machine thinks the Tyrell family is the safest in Westeros, when most of them are currently a congealed pool of goo at the bottom of a crater.