Photo on InternationalBanker.com

Anti Money Laundering

Photo on InternationalBanker.com

Anti Money Laundering

  • Worked with the largest regional bank in the South-East USA which spends a considerable amount of time and resources investigating 30k+ suspicious money laundering alerts per month to develop a model which predicts the seriousness of the alerts.
  • Led a team of 4 and tackled issues such as peer – profiling, unavailable data and false positives by feature generation and selection.
  • Tested the data with Logistic Regression, Random Forest, SVC and ADA Boost models to predict the quality of alert (High or Low) using precision and recall parameters.
  • Delivered the ADA Boost model with a precision of 38.50% and recall of 78.48% for its high recall prediction.
Python Scikit-Learn Machine Learning Logistic Regression SVM Random Forest ADA Boost Numpy Pandas Matplotlib Decision Trees SQL
Avatar
Sandeep Gunda
Aspiring Data Scientist