Dota 2 Match Outcome Prediction

Overview

This project aims to predict the outcome of Dota 2 matches based on the heroes picked by both the Radiant and Dire teams. Using machine learning techniques, we train a model to determine the likelihood of either team winning given the heroes selected during the drafting phase.

Project Structure

Data Collection: We collect match data, including the heroes picked by both teams and the match outcome, using the OpenDota API.
Data Preprocessing: The hero IDs for both teams are encoded using one-hot encoding to create a feature set for machine learning models.
Model Training: A machine learning model is trained on the preprocessed data to predict the match outcome.
Prediction: The trained model is used to predict the outcome of new matches based on the heroes picked.

Data Directory

data/
- matches.csv: Collected match data including match IDs, hero picks, and match outcomes.
- heroes.csv: List of Dota 2 heroes with their corresponding IDs.

Notes

The OpenDota API may return the same match details if queried too frequently. It is recommended to space out API requests to avoid collecting duplicate data.
The model's performance may vary depending on the amount and quality of data collected. Experimenting with different machine learning algorithms and hyperparameters is encouraged.

Future Work

Feature Engineering: Explore additional features such as player rankings, hero synergy, and counter-picks to improve model accuracy.
Model Optimization: Experiment with different machine learning algorithms and techniques like cross-validation, hyperparameter tuning, and ensemble methods.
Real-time Prediction: Integrate the model with live match data to provide real-time win probability predictions during ongoing matches.