Supervised Credit Card Fraud Detection

Data mining project exploring supervised techniques for detecting fraudulent credit card transactions.

Project Structure

• src/data/ contains load_data.py to load the dataset of credit card transactions
• src/figures/ contains all plots and images generated from the code
• src/results/ contains text files with metrics for all supervised classification models
  • Each text file contains metrics from 20 replications of running the models with the mean and standard deviation values
• src/exec_traces/ contains text files with execution traces and output from running all of the Python files in the project
• src/initial_analysis contains all code associated with the intial analysis
• src/supervised contains all code associated with the supervised classification models
  • src/supervised/ contains the logistic regression, naive Bayes, SVM, and decision tree models
  • src/supervised/ensemble_methods contains implementations of the random forest, bagging, and boosting models
  • src/supervised/neural_nets contains implementations for 2 versions of the neural network (v2 is the main neural net model)
• src/classifier_results.pdf contains the aggregated results from all of the classifiers
  • These are the results used in the paper

Usage

To run any of the Python files:

python3 [file.py] [path to creditcard.csv file]

Example:

python3 logistic_regression.py ~/Downloads/creditcard.csv