Skip to content

Detect behavioural anomalies to identify fraudulent users

Notifications You must be signed in to change notification settings

mahboobmm/fraud-detection

Repository files navigation

Fraud-Detection

Detect behavioural anomalies to identify fraudulent users

Performs One-Class Classification

Final Version.ipynb contains all the updated code

  1. Encode labels for categorical data -> Assigns numbers to categories
  2. Extracts more features from dates
  3. Merge fraudsters data with the transactions they have done to derive a pattern for each transaction
  4. Since this is one class classification, OneClassSVM is perfect for this situation, derives a pattern for fraud transactions
  5. Save all the users whose transactions were tagged by the SVM as fraudster

About

Detect behavioural anomalies to identify fraudulent users

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published