DotaMatchPrediction

This project predicts Dota 2 match outcomes using machine learning models, leveraging data from OpenDota. The workflow includes data extraction, cleaning, feature engineering, and model training (including XGBoost and linear/logistic regression).

Project Structure

.
├── .devcontainer/           # VS Code dev container config
│   ├── devcontainer.json
│   └── dockerfile
├── data/
│   ├── bronze/              # Raw data (downloaded)
│   ├── silver/              # Cleaned/transformed data
│   └── gold/                # (Reserved for final datasets)
├── src/
│   ├── fetch_data.ipynb     # Data extraction from OpenDota API
│   ├── data_wrangling.ipynb # Data cleaning & feature engineering
│   └── model.ipynb          # Model training & evaluation
├── requirements.txt         # Python dependencies
├── README.md
└── .gitignore

Setup & Installation

1. Clone the repository:

   git clone https://github.com/Tooruogata/DotaMatchPrediction.git
   cd DotaMatchPrediction

2. Set up docker image:

Set up the docker image:

docker build -t DotaMatchPrediction:latest -f .devcontainer/Dockerfile .
docker run -dit --name DotaMatchPrediction -v "$repopath:/workspace" -w /workspace DotaMatchPrediction:latest

Data Pipeline

1. Data Extraction:
   Use src/fetch_data.ipynb to download match data from OpenDota using SQL queries via their API. Data is saved in data/bronze/all_dota_matches.csv.

2. Data Wrangling:
   Clean and transform the raw data using src/data_wrangling.ipynb. This notebook performs feature engineering and outputs a processed dataset to data/silver/data_transformed.csv.

Modeling

• Model Training & Evaluation (in progress):
  src/model.ipynb contains code for:
  • Linear Regression
  • Logistic Regression
  • XGBoost Classifier (with grid search)
  • Model evaluation (accuracy, confusion matrix, classification report, plots)

Usage

1. Fetch Data:
   Run all cells in src/fetch_data.ipynb to download and save raw match data.

2. Process Data:
   Run all cells in src/data_wrangling.ipynb to clean and engineer features.

3. Train Models:
   Run all cells in src/model.ipynb to train and evaluate models.