One lucky accuracy score is not evidence. This churn classifier is evaluated the way I'd want to ...

One lucky accuracy score is not evidence. This churn classifier is evaluated the way I'd want to ...One lucky accuracy score is not evidence. This churn classifier is evaluated the way I'd want to ...

The network for creativity

Join 1.25M professional creatives like you

Connect with clients, get discovered, and run your business 100% commission-free

Creatives on Contra have earned over $150M and we are just getting started

Back to feedPost

Jigon Yoo

• Jul 28

One lucky accuracy score is not evidence. This churn classifier is evaluated the way I'd want to be audited: leakage-safe split (scaler fit on training rows only, 1,125 train / 375 held-out), a majority baseline it must beat, and five-fold cross-validation. Committed results: held-out ROC-AUC 0.7982 vs a 0.5000 majority baseline, 5-fold CV 0.819 +/- 0.0298 - beats the baseline, stable across folds. The model card also says what the model cannot tell you. The deliverable is not the classifier; it is an evaluation you can trust enough to make a decision with. Public code: github.com/jigonyoo/ml-churn-pipeline

Python Data Science

Sumit Joshi

pro

• Jul 30

Guess 216 isn't enough to crack the top 1% Aiming for 250!!!! Lesssgo

Contra Web Development frontenddeveloper

Samuel O

pro

• Jul 30

wow that's a lot keep pushing 👍

daniel linus

• Jul 29

This Windsurf AI project had all the right features but suffered from slow performance, broken components, and deployment failures. Cleaned the codebase, fixed backend integration, optimized loading speed, and delivered a scalable web application built for real users.

Low-Code/No-Code Development Windsurf backend

An .

• Jul 30

TallyTrace — verify, then export.TallyTrace is the invoice tool I kept wishing existed: one that treats uncertainty as a workflow, not a footnote. It links every extracted value to visible source evidence, routes ambiguous fields into a focused review queue, detects risky duplicates, and previews a new-file-only spreadsheet export before anything leaves the workspace.I built this six-screen Flowstep prototype around one principle: automation should make financial work easier to verify, not harder to question. The signature interaction connects a selected field directly to its location on the source invoice, while the final run receipt makes every exported row traceable.Public Flowstep prototype:https://app.flowstep.ai/file?activeFileId=18aeb4f5-8f74-4621-8ac7-9c388cc1702927-second walkthrough and public social post:https://x.com/TobiUqun/status/2082662128687661283