IPL Forecaster is an intelligent, modular cricket analytics system that uses deep learning to predict IPL match scores. What sets it apart? Users are empowered to upload any IPL-related CSV dataset — match history, player stats, ball-by-ball data, etc. — and the system intelligently interprets, processes, and uses the data to train and generate accurate score predictions.
- ✅ Upload any structured IPL-related .csv file
- 🔍 Automatic file detection (e.g., matches, deliveries, batters)
- 🧠 Deep learning-based score prediction engine
- 🧼 Smart preprocessing and feature extraction
- 📊 Performance evaluation included (MAE, prediction vs actuals)
The system accepts any IPL dataset in .csv format, including (but not limited to):
| Type | Description | Sample Columns |
|---|---|---|
| matches.csv | Match-level data | team1, team2, venue, runs, etc. |
| batters.csv | Player batting stats (seasonal) | Player Name, Team, SR, AVG, etc. |
| deliveries.csv | Ball-by-ball match data | match_id, batsman, runs, wicket, etc. |
⚙ The system auto-detects file type based on column names and processes accordingly.
Predict the final score (target_runs) of a team based on match conditions, team composition, and historical performance — dynamically using user-supplied datasets.
-
Preprocessing:
- Column detection and validation
- Team name mapping
- Feature engineering (SR, AVG, wickets, etc.)
-
Model:
- Fully connected neural network
- Built using TensorFlow/Keras
- Optimized for MAE loss
-
Evaluation:
- Mean Absolute Error (MAE)
- Sample output comparisons
- Python 3.10+
- pandas, numpy
- scikit-learn
- TensorFlow/Keras
- Matplotlib, Seaborn
- Build custom score predictors using your own IPL dataset
- Analyze fantasy league data or custom match formats
- Educational use for learning ML on sports datasets