Skip to content
All projects
Academic project

Fraud Detection and Model Benchmarking

Applied Machine Learning Project
Random ForestSVMk-NNLogistic RegressionGradient BoostingMLPROC-AUC

Technical approach

  • Random Forest
  • Support Vector Machine
  • k-nearest neighbours
  • Logistic Regression
  • Gradient Boosting
  • Multi-layer perceptron
  • Scaling
  • Outlier handling
  • Cross-validation
  • ROC-AUC
  • Confusion-matrix analysis

Limitations

  • Imbalanced-class evaluation
  • False-positive and false-negative trade-offs
  • Need for human review
  • Responsible use in risk-sensitive decisions

Visuals and artefacts

Asset → /images/projects/fraud-detection-results.png

Technologies

Random ForestSVMk-NNLogistic RegressionGradient BoostingMLPROC-AUC