Hackathon participants had to predict the victory of one or another team in the game Dota 2.
The game Dota 2 involves two sides: teams of the forces of Light (The Radiant) and the forces of Darkness (The Dire). Each team consists of 5 players - characters with their own unique abilities. Before the start of the match, each player chooses one hero out of 100 possible, and all selected team heroes must be different.
The goal of the game is to destroy the throne located in the enemy base.
Players can receive gold and experience for killing other people's heroes or other units. The accumulated experience affects the hero's level and allows him to improve his abilities.
With the accumulated gold, players buy items that improve the characteristics of their heroes or give them new abilities. During the game, teams develop their heroes, defend their part of the field and attack the enemy.
The goal of the competition is to predict the victory of the team of Light or Darkness in a fight at the moment it begins. A victory is considered to be a team receiving more gold than the opposing team. The prediction should be based on the data provided by the players, such as: the amount of gold, experience, number of killed creeps and other indicators.
The more wins you guess, the better the result!
The latest set of data was specially collected for the participants, including information on gold, experience and the number of killed creeps and other indicators of players before the start of fights in Dota 2 matches. Data files are located in the competition_data.zip archive.
To train your models in the archive you will find train.csv, containing 49 088 fights with a known answer for them. The test data test.csv, which you need to mark up yourself, contains 49 089 fights.
To work with data, a set of files in the dictionaries directory with decryption of internal codes is also attached, which will allow you to better understand the data:
Your task is to build an algorithm that predicts victory in the fight radiant_win, and always in the form of labels: 0 and 1. The work of the constructed algorithm can be checked on test data test.csv.
To make the task easier for participants, an example file with a solution was prepared, as well as a training IPython Notebook CompetitionBaseline.ipynb with a simple but complete solution to the problem.