alibekk93 / fm_skills_prediction

The Soccer Stats Spectacular

Table of Contents

1. Project Overview
2. Data Preprocessing
3. Installation
4. Usage
5. Contribution
6. License
7. Contact

Project Overview

Welcome to the Soccer Stats Spectacular! This project dives into the world of soccer to analyze player performance and predict their skills. We're combining real-world match statistics with player ratings from Football Manager 2023 and FC24 to create a comprehensive model for player evaluation and comparison.

What’s the Game Plan?

1. Predict Player Skills: Can we use real match data to predict how players are rated in video games? Let's find out!
2. Find Player and Team Twins: Ever wondered which players or teams are secretly alike? We'll uncover those hidden similarities.
3. Bridge the Gap: Explore the fascinating relationship between on-field performance and virtual ratings.

Get ready for a deep dive into soccer stats, where data science meets the beautiful game!

Data Preprocessing

1. Data Consolidation 📓

• Merge separate files for each league and statistic type.
• Combine FBRef statistics with Football Manager and FC24 ratings.
• Ensure consistent player and team naming across all datasets.

2. Feature Engineering

• Calculate percentage above/below team average for each player statistic:

  def percent_above_below(group):
      return (group - group.mean()) / group.mean()
  
  player_data_transformed = player_data.groupby('team').  transform(percent_above_below)

• Aggregate original pre-transformation data per team for team-level statistics.

3. Dimensionality Reduction for Player Statistics

• Apply Principal Component Analysis (PCA) to reduce the dimensionality of player statistics:

  from sklearn.decomposition import PCA
  from sklearn.preprocessing import StandardScaler
  
  scaler = StandardScaler()
  player_data_scaled = scaler.fit_transform(player_data_transformed)
  
  pca_player = PCA(n_components=0.95)  # Retain 95% of variance
  player_pca = pca_player.fit_transform(player_data_scaled)

4. Team Style and Strength Encoding

• Apply PCA to reduce the dimensionality of team statistics.
• Append the resulting team PCA components to individual player data:

  team_data_scaled = scaler.fit_transform(team_data)
  pca_team = PCA(n_components=5)  # Adjust number of components as needed
  team_pca = pca_team.fit_transform(team_data_scaled)
  
  player_data_final = pd.concat([
      pd.DataFrame(player_pca),
      pd.DataFrame(team_pca[player_data['team_id']])
      ], axis=1)

5. Data Validation and Exploratory Analysis

• Perform exploratory data analysis (EDA) after each preprocessing step.
• Visualize PCA results to understand player and team distributions.
• Validate the effectiveness of the preprocessing steps.

6. Data Versioning

• Implement a versioning system for datasets at each preprocessing stage.
• Ensure reproducibility of results and enable easy backtracking if needed.

Installation

To get started with the Soccer Stats Spectacular, follow these steps:

1. Clone the repository:

git clone https://github.com/your-username/soccer-stats-spectacular.git

2. Navigate to the project directory:

cd soccer-stats-spectacular

3. Install the required packages:

pip install -r requirements.txt

Usage

To run the analysis and start making predictions, execute the following command:

python main.py --data-source "FBRef" --season "2023-24" --predict-skills

This will kick off the data processing and prediction pipeline, producing results that you can analyze further.

Contribution

We welcome contributions from fellow soccer enthusiasts and data scientists! To contribute:

1. Fork the repository.
2. Create your feature branch (git checkout -b feature/AmazingFeature).
3. Commit your changes (git commit -m 'Add some AmazingFeature').
4. Push to the branch (git push origin feature/AmazingFeature).
5. Open a Pull Request.

License

This project is licensed under the MIT License. See the LICENSE.md file for details.

Contact

For questions, support, or just to say hello, reach out to us at:

• Email: [email protected]
• Twitter: @SoccerStatSpectacular

We hope you enjoy diving into the world of soccer stats with us!