https://github.com/mehrab-kalantari/credit-card-fraud-detection

Credit card transactions fraud detection using classic algorithms
https://github.com/mehrab-kalantari/credit-card-fraud-detection

association-analysis auc-roc-curve correlation-analysis credit-card-fraud-detection feature-engineering fraud-detection imbalanced-learning model-selection roc-curve smote-tomek

Last synced: about 2 months ago
JSON representation

Credit card transactions fraud detection using classic algorithms

• Host: GitHub
• URL: https://github.com/mehrab-kalantari/credit-card-fraud-detection
• Owner: Mehrab-Kalantari
• Created: 2024-01-18T14:08:15.000Z (over 1 year ago)
• Default Branch: main
• Last Pushed: 2024-01-18T14:32:13.000Z (over 1 year ago)
• Last Synced: 2025-01-16T09:42:28.675Z (4 months ago)
• Topics: association-analysis, auc-roc-curve, correlation-analysis, credit-card-fraud-detection, feature-engineering, fraud-detection, imbalanced-learning, model-selection, roc-curve, smote-tomek
• Language: Jupyter Notebook
• Homepage:
• Size: 2.07 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Credit Card Transactions Fraud Detection

[Dataset on kaggle](https://www.kaggle.com/datasets/kartik2112/fraud-detection)

## Contents

### Data Understanding

* Features

* Null values detection

* Duplicated values detection 

### Data Cleaning For EDA

* Column removal

* Discretization

* Creating new features

* Feature extraction

### Feature Engineering

* Datetime feature extraction

* Credit card feature extraction

### Exploratory Data Analysis

* Univariate Analysis

  * Target

  * Categorical features

  * Numerical features

* Bivariate Analysis

  * Target analysis

  * Amount of activity analysis

  * Time analysis

### Correlation and Association Analysis

* Correlation matrix 

* Association matrix

### Data Preprocessing

* Column removal

* Log transform

* Categorical encoding

  * Binary encoding

  * Weight of evidence encoding

  * Ordinal encoding

* Train-test split

### Imbalanced Learning

Target is imbalanced

* ![p](sample/target.png)

Methods performed

* No changes

* Random under sampling

* Random over sampling

* SMOTE-Tomek links

* Class weights

### Feature Importance

![pp](sample/feature%20importance.png)

### Modeling

1. Random Forest Classifier

2. Logistic Regression Classifier

3. Naive Bayes

4. Decision Tree Classifier

5. Support Vector Machine (SVM) Classifier

6. K-nearest neighbor (KNN) Classifier

### Evaluation

* Confusion matrix

* AUC curve

* Classification metrics

* Decision boundary

Results on random forest classifier for test data

* ![p2](sample/conf.png)

* ![p3](sample/roc.png)

### Model Selection

Results on all models for test data

![all](sample/selection.png)