This project is about solving the binary classification problem (fraud detection) based on the dataset retrieved from https://www.kaggle.com/mlg-ulb/creditcardfraud. Please check the paper for more details.

Kirby Zhou, Wenhao Su, James Lemkin, Zekun Chen, Zhujun Fang,Yuqi Sha,

Rongfei Li, Hulin Wang, Damu Gao, Bo Xiao, Haoyang Li, Yizhi Huang

The file folder libs consists of multiple utility functions to draw the ROC and PR curves. Before running the script, it is extremely important to download corresponding processed Data folder AND dataset from Kaggle website to current workspace.

Another repository is merged into current repository: https://github.com/cosmobiosis/fraud_detection

Dataset Folder download link: https://github.com/DeviRule/Ecs171_group-project/tree/master/Data

Kaggle Dataset download link: https://www.kaggle.com/mlg-ulb/creditcardfraud

The output will be automatically saved to current folder or printed out during runtime.

For decision tree script, please extract code to parent directory.

The project consists of file scripts, together with utility functions inside the file folder Utility.

decision_tree_grid_search.py: decision tree

distribution.py: preprocessing distribution

preprocessing_comparision.ipynb: SMOTE performance

Adaboost_main.py: Adaboost

SNE_KNN.ipynb: TSNE and KNN

ANN_kfold.ipynb: Feed-Forward Neural Network

RF_kfold_under.ipynb: Random Forest ROC graph

RF_final_under.ipynb: Random Forest Pr Curve, Classification Report

logistic_regression_parameter_sweep_undersample.ipynb: Logistic Regression Parameter Tuning

logistic_regression_kfold.py: Logistic Regression Best Model Performance

• Programming Tools: Python3 and Jupyter Notebook
• Libraries: imblearn, sklearn, Keras, matplotlib, graphviz and pandas

In electronic pdf paper file, every link is connected to the corresponding script file which can retrieve and reproduce the results from the paper.

For any questions contact [email protected] (email expired in 2021).

The project is using MIT license.

This work was supported by course ECS 171 of UC Davis. The dataset has been collected and analysed during a research collaboration of Worldline and the Machine Learning Group (http://mlg.ulb.ac.be) of ULB (Université Libre de Bruxelles) on big data mining and fraud detection. More details on current and past projects on related topics are available on https://www.researchgate.net/project/Fraud-detection-5 and the page of the DefeatFraud project