IEEE Fraud Detection with Anomaly Detection Algorithms

IEEE Fraud Detection is a competition that you are predicting the probability that an online transaction is fraudulent, as denoted by the binary target isFraud. The data is broken into two files identity and transaction, which are joined by TransactionID. Not all transactions have corresponding identity information.

This repo provides training hyperparameter optimization and evaluation scripts for IEEE Fraud Detection data with XGBoost and CatBoost algorithms.

Inspect CatBoost notebook and XGBoost notebook for detailed demo.

Usage

• Clone:

git clone https://github.com/fcakyon/ieee-fraud-detection.git

• Prepare conda environment:

conda env create -f environment.yml

conda activate frauddetection

• Fir XGBoost or CatBoost model:

python train_xgboost.py

python train_catboost.py

Detailed Usage

• After downloading the data to ./data folder, you can get the preprocessed data via get_data() method in preprocessing.py module:

from preprocessing import get_data

float: val_split = 0.2 # splits %80 of data for train and %20 for val
bool: apply_label_encoding = True # applies label encoding to categorical features
bool: fillna = True # fills missing values with -999

data = get_data(val_split, apply_label_encoding, fillna)

• After initializing CatBoost or XGBoost classifier, perform automatic hyperparameter optimization via perform_random_search() or perform_bayes_search() in hyperparam_optimizing.py module:

from hyperparam_optimizing import perform_bayes_search, CATBOOST_BAYESSEARCH_PARAMS
import catboost as cb

# define xgboost or catboost instance
estimator = cb.CatBoostClassifier(
    n_estimators=200,
    learning_rate=0.05,
    metric_period=500,
    od_wait=500,
    task_type="CPU",
    depth=8,
) 

# parse get_data() output
X_train = data["X_train"]
X_val = data["X_val"]
y_train = data["y_train"]
y_val = data["y_val"]

# define parameter grid
param_grid = CATBOOST_BAYESSEARCH_PARAMS

# define scoring metric, full list can be seen at https://scikit-learn.org/stable/modules/model_evaluation.html
scoring = 'roc_auc'

# perform bayes parameter search for catboost classifier
best_estimator = perform_bayes_search(
    estimator, X_train, X_val, y_train, y_val, param_grid, scoring
)

• After fitting a model, print results such as accuracy, auc score, confusion matrix, f1 scores using calculate_scores() method from scoring.py module:

from scoring import calculate_scores

scores = calculate_scores(estimator=best_estimator, X_val=X_val, y_val=y_val)