What I Learned Analyzing 20,000 Airbnb ListingsMy first market analysis: capping price outliers, why location matters at neighbourhood not borough level.5 min read·beginnerslog.hashnode.dev12SOct 15
My Final Fraud Model (Day 7)Test AUC was 0.9571, lower than validation's 0.9765. Here's why the test number is the honest one, and why threshold mattered more than model choice.5 min read·beginnerslog.hashnode.dev1Oct 8
Picking a Threshold Is a Business Decision (Day 6)The chart doesn't decide for you. I had to.4 min read·beginnerslog.hashnode.devOct 4
XGBoost Caught More Fraud - But There's a Catch (Day 5)My recall jumped from 59% to 81%. My precision dropped from 79% to 57%.4 min read·beginnerslog.hashnode.devOct 4
I Finally Understand AUC, Confusion Matrix, and Classification Report (Day 4)My first fraud detection model scored 0.9571 AUC and 99.9% accuracy but missed 40% of frauds.4 min read·beginnerslog.hashnode.devSep 29
