🏅 Olympic Medal Prediction: 2028 Los Angeles Games

A comprehensive machine learning system utilizing Gradient Boosting Regression Trees (GBRT) to analyze historical Olympic data and predict medal counts for the 2028 Los Angeles Summer Olympics.

Based on the research paper: Prediction and Analysis Based on the GBRT Model.

---

📊 Key Visualizations

💡 Note: These plots are generated by running the visualization scripts in src/visualization/.

2028 Medal Predictions (Total)

Forecasted total medal counts for top performing nations, including confidence intervals.

Feature Importance Analysis

Analysis of factors contributing to Gold vs. Total medal counts.

📑 Research Insights

Derived from the core GBRT analysis in our Research Paper.

1. GBRT Model Framework

A systematic approach to multi-feature regression for Olympic medal forecasting.

2. 2028 Medal Prediction World Map

Global heatmap showing predicted gold medal distribution for the 2028 Olympics.

3. Winning Rate Correlation Heatmap

Heatmap showing the winning rates of top 10 countries across different events.

4. Model Residuals Analysis

Residual plots validating the GBRT model's prediction accuracy for gold and total medals.

---

🚀 Key Features

• Advanced Data Cleaning: Robust pipelines to handle multi-source data (Athletes, Medals, Hosts, Programs).
• Feature Engineering: Extraction of critical signals including Host Country Effect and Historical Momentum.
• GBRT Modeling: Optimized Gradient Boosting Regressor with Grid Search hyperparameter tuning.

---

📂 Project Structure

Predicting-Medals/
├── data/
│   ├── raw/                 # Original datasets (from Kaggle/Olympic.org)
│   └── processed/           # Cleaned and merged CSVs
├── docs/                    # Research papers and documentation
├── outputs/                 # Generated plots and prediction results
├── result/                  # Static visualization assets from research paper
├── src/
│   ├── analysis/            # Statistical analysis scripts (Coach effect, etc.)
│   ├── data_cleaning/       # Preprocessing pipelines
│   ├── feature_engineering/ # Feature construction
│   ├── models/              # GBRT and Linear Regression models
│   └── visualization/       # Plotting scripts
├── requirements.txt         # Project dependencies
├── .gitignore               # Git exclusion rules
├── LICENSE                  # MIT License
└── README.md

---

🛠️ Installation

1. Clone the repository

   git clone https://github.com/1EchA/Predicting-medals.git
   cd Predicting-medals

2. Install dependencies

   pip install -r requirements.txt

3. Prepare Data Unzip the raw data into the data/raw directory:

   unzip data/Data.zip -d data/raw/

---

💻 Usage

To reproduce the analysis and predictions, follow this pipeline:

1. Data Cleaning

Standardize names, handle missing values, and merge datasets.

python src/data_cleaning/clean_athletes.py
python src/data_cleaning/clean_medals.py
python src/data_cleaning/clean_hosts.py
python src/data_cleaning/clean_programs.py

2. Feature Engineering

Construct the training dataset with historical features.

python src/feature_engineering/build_dataset.py
python src/feature_engineering/merge_events.py

3. Model Training & Prediction

Train the GBRT model and generate predictions for 2028.

python src/models/gbrt_model.py

4. Visualization

Generate the plots shown above.

python src/visualization/gbrt_visualization.py
python src/visualization/feature_visualization.py

---

📈 Model Performance

The GBRT model was tuned using Grid Search with 5-fold cross-validation.

Metric	Value
Model	Gradient Boosting Regressor
CV MSE	~Optimized
Key Hyperparams	n_estimators: [100, 200], learning_rate: [0.05, 0.1]

Key Findings:

• Host Effect: Significant boost in medal counts for host nations.
• Historical Momentum: Previous games' performance is the strongest predictor.
• Gender Parity: Balanced gender ratios correlate with higher overall medal counts in modern games.

---

👥 Authors

• 1EchA - Lead Developer & Researcher

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.