Real-time cricket match outcome prediction with interactive analytics dashboard
🎯 Real-time Predictions · Get instant win probability predictions during live IPL matches
📊 Interactive Dashboard · Clean and intuitive Streamlit web interface with visual analytics
🏆 Team Analysis · Support for all 8 major IPL teams with historical data insights
📈 Match Progression · Track probability changes throughout the match with visual indicators
🎨 Dynamic Visualization · Progress bars and metrics showing current match situation
🔍 Detailed Analytics · Required run rate, wickets remaining, and match state analysis
Python · Core programming language
Streamlit · Web application framework
Pandas & NumPy · Data manipulation and numerical analysis
Scikit-learn · Machine learning algorithms and preprocessing
Logistic Regression · Primary prediction model
Pickle · Model serialization and deployment
CSS/HTML · Custom styling for enhanced UI
Prerequisites
Python 3.7+ · pip package manager
Installation
# Clone the repository
git clone https://github.com/svdexe/IPL_Win_Probability_Prediction.git
cd IPL_Win_Probability_Prediction
# Install dependencies
pip install streamlit pandas scikit-learn numpy pickle-mixin
# Run the application
streamlit run app.pyOpen your browser and navigate to http://localhost:8501
IPL_Win_Probability_Prediction/
│
├── app.py # Main Streamlit application
├── IPL_Win_Probability_Predictor.ipynb # Jupyter notebook with model development
├── pipe.pkl # Trained ML pipeline
├── matches.csv # IPL matches dataset
├── deliveries.csv # Ball-by-ball delivery data
├── requirements.txt # Python dependencies
├── README.md # Project documentation
└── Predictio_Website_Screen.png # Interface screenshot
Data Processing · Merges IPL matches and deliveries datasets · Filters second innings data for chase scenarios · Handles team name standardization and data cleaning
Feature Engineering · Calculates current run rate (CRR) and required run rate (RRR) · Tracks wickets remaining and balls left · Creates situational features like runs needed and match pressure
Model Training · Uses Logistic Regression with OneHotEncoder for categorical features · Trains on historical IPL data with 80-20 train-test split · Achieves high accuracy in predicting match outcomes
Real-time Prediction · Takes current match state as input · Processes through trained ML pipeline · Returns win probability for both teams with confidence intervals
The prediction system uses Supervised Machine Learning
Model · Logistic Regression with liblinear solver
Features · Batting team, bowling team, city, runs left, balls left, wickets remaining, target score, CRR, RRR
Preprocessing · OneHotEncoding for categorical variables, feature scaling
Evaluation · Accuracy score, probability calibration
Source · IPL Historical Data (2008-2020+)
Matches · 800+ IPL matches
Deliveries · 150,000+ ball-by-ball records
Teams · 8 major IPL franchises
Features · Match details, player statistics, venue information
- Select Teams · Choose batting and bowling teams from dropdown
- Set Match Context · Select host city and target score
- Enter Current State · Input current score, overs completed, and wickets lost
- Get Prediction · Click predict to see win probabilities and match insights
- Analyze Results · View detailed breakdown with required rate and match pressure
- Live API Integration · Real-time data from cricket APIs
- Player Impact · Individual player performance metrics
- Weather Conditions · Include weather impact on match outcomes
- Venue Analysis · Stadium-specific historical performance
- Mobile App · Native mobile application for on-the-go predictions
- Advanced Models · Neural networks and ensemble methods
- Historical Comparison · Compare current match with similar past scenarios
Accuracy · 85%+ prediction accuracy on test data
Precision · High precision for close match scenarios
Recall · Effective identification of winning probabilities
F1-Score · Balanced performance across different match situations
This project is licensed under the MIT License - see the LICENSE file for details.
CampusX · for the comprehensive original tutorial that served as the foundation for this project
IPL Data · Historical match and delivery data from official sources
Streamlit · Amazing framework for rapid web app development
Cricket Analytics Community · Inspiration from cricket data science projects
Shivam Dali · shivamdali@gmail.com
GitHub · https://github.com/svdexe
LinkedIn · https://www.linkedin.com/in/shivam-dali-86b0a1201/
Project Link: https://github.com/svdexe/IPL_Win_Probability_Prediction
⭐ Star this repo if you found it helpful!
Made with ❤️ and lots of ☕