ImpactMeter

ImpactMeter is a context-aware cricket analytics system that converts ball-by-ball IPL data into an interpretable Impact Metric (IM_score) ranging from 0–100. It evaluates player performance using match context, pressure situations, and recent form.

Live Demo:

https://mr-baraiya.github.io/ImpactMeter

---

Why ImpactMeter

Traditional cricket metrics such as runs, strike rate, or wickets fail to capture when and how important a performance was.

ImpactMeter improves this by modeling:

• Performance – runs, strike rate, batter/bowler impact
• Context – powerplay, middle overs, death overs
• Situation – pressure-aware contribution and clutch performance
• Recency – rolling and weighted recent-match form

Core idea:

Impact = Performance × Context × Pressure

This produces a normalized Impact Score (0–100).

---

Requirements

• Python 3.10+
• pip
• A local web server (Python http.server works)

---

Project Structure

ImpactMeter/
│
├── index.html
├── script.js
├── requirements.txt
│
├── data/
│   ├── raw/ipl_json/
│   ├── processed/ball_by_ball.csv
│   └── features/
│       ├── impact_dataset.csv
│       └── player_impact_scores.csv
│
├── models/
│   ├── ml_impact_scores.csv
│   └── ml_feature_importance.csv
│
├── scripts/
│   ├── json_to_csv.py
│   ├── run_impact_model.py
│   └── run_ml_assisted_impact.py
│
├── notebooks/
│   ├── analysis.ipynb
│   ├── feature_engineering.ipynb
│   ├── impact_model.ipynb
│   └── ml_assisted_impact.ipynb
│
├── docs/
│   ├── 01_problem_definition.md
│   ├── 02_dataset.md
│   ├── 03_feature_engineering.md
│   ├── 04_impact_model.md
│   ├── 05_algorithm_pipeline.md
│   ├── 06_results.md
│   ├── 07_edge_cases.md
│   └── 08_impact_metric_explanation.md
│
└── frontend/
    └── ImpactMeter_Logic_Design_Document.html

---

Setup

Run from project root:

python -m venv venv
.\venv\Scripts\Activate
python -m pip install --upgrade pip
pip install -r requirements.txt

---

Quickstart (End-to-End)

To rebuild the full pipeline:

# Convert Cricsheet JSON → ball-by-ball dataset
python scripts/json_to_csv.py

# Run feature engineering notebook
notebooks/feature_engineering.ipynb

# Generate rule-based impact scores
python scripts/run_impact_model.py

# Generate ML-assisted validation scores
python scripts/run_ml_assisted_impact.py

# Launch dashboard
python -m http.server 8000

Open:

http://localhost:8000

---

Detailed Pipeline

1. Convert Raw JSON

python scripts/json_to_csv.py

Output:

data/processed/ball_by_ball.csv

---

2. Feature Engineering

Run:

notebooks/feature_engineering.ipynb

Output:

data/features/impact_dataset.csv

Features include:

• strike_rate
• phase (powerplay / middle / death)
• pressure_index
• batter_impact_score
• bowler_impact_score

---

3. Build Rule-Based Impact Scores

python scripts/run_impact_model.py

Output:

data/features/player_impact_scores.csv

Main columns:

• player
• match_id
• match_date
• impact_score
• rolling_IM
• weighted_IM
• pressure_score
• clutch_score
• IM_score

---

4. ML-Assisted Impact Validation

python scripts/run_ml_assisted_impact.py

Outputs:

models/ml_impact_scores.csv
models/ml_feature_importance.csv

ML is used for:

• feature importance validation
• score stability verification

Rule-based scoring remains the primary metric.

---

Dashboard

Start server:

python -m http.server 8000

Open:

http://localhost:8000

Files used:

• UI → index.html
• Logic → script.js

---

Documentation

Main project documentation:

• Problem Definition → docs/01_problem_definition.md
• Dataset → docs/02_dataset.md
• Feature Engineering → docs/03_feature_engineering.md
• Impact Model → docs/04_impact_model.md
• Algorithm Pipeline → docs/05_algorithm_pipeline.md
• Results → docs/06_results.md
• Edge Cases → docs/07_edge_cases.md
• Impact Metric Explanation → docs/08_impact_metric_explanation.md
• Technical Design → frontend/ImpactMeter_Logic_Design_Document.html

---

Troubleshooting

If CSV files fail to load in the browser:

• ensure the server was started from repo root

If ML outputs are missing:

python scripts/run_ml_assisted_impact.py

If rule scores are missing:

python scripts/run_impact_model.py

---

Data Source

Cricsheet IPL dataset https://cricsheet.org/