World of Warcraft is a massively online multiplayer role playing game (MMORPG). One of the features of the game are these scripted boss encounters where anywhere from 10 to 40 players come together all to kill a single or series of bosses in the game. These encounters can take a lot of coordination, preparation and individual player skill to successfully defeat the boss.

Whenever a new encounter comes out, there is usually a pretty big world wide race to see who can kill all the new bosses first. Most notably the guild Method achieved this recently with their world first kill of Jaina in the raid Battle of Dazar’alor on the highest difficulty.

I watched some of the event and it got me interested in World of Warcraft again but from a data and analysis point of view.

What attributes about a player group lead to a successful outcome in World of Warcraft?

There are the obvious parts to it like making sure players have the best gear, the correct buffs applied, using potions, correct player composition etc. But what other aspects are there besides the obvious that correlate to successful boss kills? Can these data points be applied to predict an outcome for pugs** as well?

Thanks to community driven sites like WoW Progress, Warcraft Logs, Raider IO, and the Blizzard API, there is plenty of data available in order to begin exploring this.

In future posts, I'll discuss collection and exploration of the data to see if any specific data points have a correlation to success or failure. Then using this data, use it to predict live groups within the game itself.

** PUGS = Pick up group. Strangers coming together to achieve a goal.