FS
FraudShield
model ready · random forest

Fill in the details and run analysis

1,248
Analyzed
37
Frauds blocked
₹4.2L
Amount saved
0.974
ROC-AUC

IDCardAmountCategoryHourScoreStatus

Random Forest — 100 decision trees vote on each transaction. Majority wins. Doesn't overfit like a single decision tree would.

Fraud is rare (~3%). SMOTE creates synthetic fraud samples during training so the model doesn't just say "always legit."

Raw data
Feature engineering
Train/test split
StandardScaler
SMOTE
Random Forest
Probability score
Decision (≥0.5 = fraud)

Built by

The team behind FraudShield — SKIT Jaipur, CSE Batch 2023