IPL Pulse Lab

IPL Pulse Lab is a full-stack IPL analytics and forecasting platform focused on season intelligence, model evaluation, and report-ready outputs for academic and portfolio use.

Current Scope

• IPL-first data model (teams, players, fixtures, matches, player-match stats)
• Official-style data ingestion pipeline for:
  • season player stats
  • upcoming fixtures
  • points table override
• Analytics views for teams and players with role-aware filtering
• Prediction module:
  • fixture-based match prediction
  • run forecast (batters, wk-batters, all-rounders)
  • wicket forecast (bowlers, all-rounders)
• Evaluation module:
  • rolling time-split classification evaluation
  • calibration curve and Brier score
  • rolling one-step regression evaluation for runs/wickets
  • uncertainty bands and role-wise error breakdown
• Reporting exports:
  • player CSV
  • evaluation JSON/CSV/Markdown
  • evaluation ZIP bundle (all formats)

Tech Stack

• Backend: Flask, Flask-CORS, SQLite, NumPy, scikit-learn
• Frontend: React, Recharts
• Data: SQL schema + seed + CSV ingestion pipeline

Project Structure

backend/
	app.py
	analytics/
	database/
	models/
	routes/
	utils/
frontend/
	public/
	src/
database/
	schema.sql
	insert_demo_data.sql
	sample_queries.sql
data/
	external/
	processed/
	raw/
notebooks/
	data_cleaning.ipynb
	eda_analysis.ipynb
	model_building.ipynb
scripts/
tests/

Local Setup

Backend

pip install -r requirements.txt
cd backend
python app.py

Backend runs on http://localhost:5000.

Frontend

cd frontend
npm install
npm start

Frontend runs on http://localhost:3000.

If required, set REACT_APP_API_URL to point to a different backend.

Useful Scripts (PowerShell)

./scripts/start-backend.ps1
./scripts/start-backend-prod.ps1
./scripts/start-frontend.ps1
./scripts/health-check.ps1
./scripts/verify-all.ps1

Key API Endpoints

Health and Core Analytics

• GET /api/health
• GET /api/analytics/overview
• GET /api/analytics/evaluation

Entity Data

• GET /api/players
• GET /api/players/:id
• GET /api/teams
• GET /api/teams/:id
• GET /api/teams/points-table?season=2026

Prediction APIs

• GET /api/predictions/upcoming-matches?season=2026
• POST /api/predictions/upcoming-match
• POST /api/predictions/match
• GET /api/predictions/player/:player_id?next_matches=5
• GET /api/predictions/player-wickets/:player_id?next_matches=5

Admin/Data Refresh

• POST /api/admin/reinitialize
• POST /api/admin/ingest-live-data
• POST /api/admin/ingest-player-stats
• POST /api/admin/refresh-official-data

Report Exports

• GET /api/reports/players.csv
• GET /api/reports/evaluation.json
• GET /api/reports/evaluation.csv
• GET /api/reports/evaluation.md
• GET /api/reports/evaluation.zip

Evaluation Notes

• Match evaluation uses rolling time split to avoid temporal leakage.
• Wicket fair-comparison metrics are role-qualified:
  • roles: Bowler + All-Rounder
  • preferred minimum innings per player: 5
  • auto-fallback to 3 if preferred threshold has zero samples

Data Files and Ingestion

Official Points Table Override

Use data/external/ipl_points_2026_official.csv with columns:

• position, team, played, won, lost, no_result, points, nrr

Player Season Stats

Use data/external/ipl_player_stats_2026.csv with columns:

• full_name, team, matches_played, runs, wickets, strike_rate, economy
• optional: role

Upcoming Fixtures

Use data/external/ipl_upcoming_matches_2026_official.csv for fixture-linked predictions.

Tests

Run regression tests:

./.venv/Scripts/python.exe -m unittest tests/test_api.py -v

CI

GitHub Actions workflow in .github/workflows/ci.yml runs backend checks and frontend build.