Research in Fraud Detection using Data Mining

Support: contains some plots made for analysis of both the Paysim dataset and the final result.

Titanic Dataset: before diving into paysim analysis, I did a study with the Titanic dataset.

SMOTE: notebook on how to apply smote to supervised classification learning.

PaySim Dataset: analysis of a fictitious dataset, available on Kaggle, of mobile financial transactions that contains more than 6 thousand labels, 11 columns, no data missing, and approximately 4000 fradulent transactions. I did an exploratory analysis of the data, then I balanced the dataset to apply the classifiers. The best performing classifiers were the ensemble ones.