Predict PUBG Placement Percentage using Machine Learning

Information about the game

PUBG: Battlegrounds (previously known as PlayerUnknown's Battlegrounds) is a battle royale game developed and published by PUBG Studios, a subsidiary of Krafton. The game, which was inspired by the 2000 Japanese film Battle Royale, is based on previous mods created by Brendan "PlayerUnknown" Greene for other games, and expanded into a standalone game under Greene's creative direction. In the game, up to one hundred players parachute onto an island where they scavenge for weapons and equipment to kill other players while avoiding getting killed themselves. The available safe area of the game's map decreases in size over time, directing surviving players into an ever tightening space to force encounters.

Players can also choose to team up with friends or random players in group of two or four.The dataset contains metrics of various players we will use to train ML models and will predict the win percentage of the respective players.

Evaluation Metric :mean absolute error

In this project, MAE is used as evaluation metric.Absolute error refers to the magnitude of difference between the prediction of an observation and the true value of that observation. MAE takes the average of absolute errors for a group of predictions and observations as a measurement of the magnitude of errors for the entire group.The expression is given below.

Project Steps

In this project, we will use regression model like linear regression and decision tree models like Random forest,XGboost and LightGBM to predict the win placement.We will deal with Supervised Machine learning methods.

• Download the dataset from Kaggle
• Basic Understanding of Data & perform EDA.
• Create/prepare datasets for models
• Train Hard-coded Model
• Train various Baseline Models
• Perfrom Feature engineering
• Train and Evaluate LightGBM
• Hyperparameter Tuning
• Predictions on test Data
• Submission in Kaggle