rohit jishtu

IPL Crunch: Data Over Opinions

IPL Crunch ’26 asked us to work with real ball-by-ball data and answer high-impact questions: • Do teams that win the toss actually win more matches? • Which phase matters most — Powerplay, Middle, or Death? • Who are the top batters and bowlers in terms of winning, not just career stats? • What hidden patterns change how we read IPL results? Process I treated this like a real analytics project, not a one-off notebook. 1. Understand the brief — mapped Wooble’s questions to measurable metrics. 2. Audit the data — inspected columns, match counts, team/venue label inconsistencies. 3. Clean & normalize — merged duplicate franchise and stadium names before aggregating. 4. Engineer metrics — built win-contribution ratings, chase/defense scores, MVP scoring, and win-driver tags. 5. Analyze in Python — single-pass pipeline over ~289K ball rows. 6. Tell the story — written report + interactive dashboard with filters. 7. Ship & validate — rebuilt artifacts, published repo + live GitHub Pages dashboard. Tools: Python 3 (stdlib only), CSV processing, static HTML dashboard with client-side filtering. No black-box ML Results Live: https://rohitjishtu.github.io/ipl-crunch-26/dashboard/index.html Reflection Phase filters in-browser · lighter deploy (dashboard only) · test one stranger on filters before submit.