Skip to content

README.md

Latest commit

 

History

History
105 lines (76 loc) · 2.36 KB

README.md

File metadata and controls

105 lines (76 loc) · 2.36 KB
Raw

FIFA Player Price Prediction

Machine learning project for predicting football player market value in EUR from tournament performance, physical attributes, position, and match statistics.

The project contains two baseline models:

  • RandomForestRegressor in src/train.py
  • GradientBoostingRegressor in src/train1.py

Current local results:

Model MAE R2
Baseline mean prediction 17,635,635 -
Random Forest 6,266,188 0.761
Gradient Boosting 8,768,228 0.664

Project Structure

FIFAPlayerPrices/
├── data/
│   ├── raw/
│   │   └── player_performance.csv
│   └── processed/
│       └── players_clean.csv                 # generated locally
├── models/
│   ├── player_price_model.pkl                # generated locally
│   └── gradient_boosting_player_price_model.pkl # generated locally
├── notebook/
│   └── fifa_player_price_prediction.ipynb
├── src/
│   ├── features.py
│   ├── paths.py
│   ├── prepare_data.py
│   ├── predict.py
│   ├── train.py
│   └── train1.py
├── .gitignore
├── README.md
└── requirements.txt

Generated processed data and model files are ignored by Git because they can be recreated locally and model files can be large.

Setup

Create and activate a virtual environment:

python3 -m venv .venv
source .venv/bin/activate

Install dependencies:

pip install -r requirements.txt

Usage

Prepare the cleaned dataset:

python src/prepare_data.py

Train the Random Forest model:

python src/train.py

Train the Gradient Boosting model:

python src/train1.py

Run example predictions with both trained models:

python src/predict.py

Features

The model uses:

  • player position
  • age, height, and weight
  • shooting, passing, dribbling, defensive, and goalkeeper statistics
  • physical metrics such as distance covered, sprint distance, top speed, stamina
  • tournament-level performance metrics

The target column is:

market_value_eur

Notes

The Random Forest model currently performs better than Gradient Boosting on the same train/test split. A good next improvement would be to train on log(market_value_eur) because football market values are usually highly skewed.