Credit Card Fraud Detection with Machine Learning

This repository contains the code for a credit card fraud detection system using machine learning techniques. The project is designed to help identify fraudulent transactions in real-time, minimizing financial losses and protecting consumers.

Goal

We aim to build a model to detect if a transaction is normal or fraudulent using machine learning techniques.

Structure of project

1. Importing Libraries

   • Pandas & NumPy: For data manipulation.
   • Matplotlib & Seaborn: For visualization.
   • Scikit-learn: For model building (Logistic Regression, Decision Tree, Random Forest) and performance metrics.
   • Imbalanced-learn (SMOTE): To handle imbalanced datasets.
   • Joblib: For model serialization.

2. Loading and Exploring the Dataset

   • Data Source: The dataset is available on Kaggle, containing details of transactions labeled as normal or fraudulent.
   • Data Cleaning:
   • Duplicate data is removed.
   • Null values are confirmed as absent.

3. Data Visualization

   • Histograms and box plots are used to understand the distribution of transaction amounts.
   • Count plot displays the distribution between fraudulent and non-fraudulent transactions.

4. Handling Class Imbalance

   • The dataset has a class imbalance, with far fewer fraudulent transactions.
   • Class Imbalance Handling:
     • Undersampling: Reducing the majority class size.
     • SMOTE: Synthetic data generation for the minority class.

5. Feature Engineering

   • The Amount column is standardized to match the range of other features.
   • The Time column is dropped as it's irrelevant to the classification task.

6. Model Selection

• Three models are evaluated:
  • Logistic Regression
  • Decision Tree
  • Random Forest
• Performance metrics include:
  • Accuracy: Percentage of correct predictions.
  • Precision: How many selected items are relevant.
  • Recall: How many relevant items are selected.
  • F1-Score: Harmonic mean of precision and recall.
  • Confusion Matrix: Evaluates true positives, true negatives, false positives, and false negatives.

7. Final Model - Random Forest
   • Random Forest achieved the best performance with an accuracy of 99.99%.
   • The model is saved as CerditCardRandomForestModel.pkl using joblib for later use.
8. Deploying the Model
   • The trained Random Forest model can be loaded and tested on new data to predict if a transaction is fraudulent.

Conclusion

• Random Forest is chosen as the final model due to its superior accuracy, recall, and precision.
• This solution is highly applicable in real-world settings like fraud detection where minimizing both false positives and negatives is critical.

Links

• Kaggle Notebook
• LinkedIn Account
• Email

how to run project >>

1. Clone the repository:

    git clone https://github.com/MuhammadRuby/Crdit_Card_Fraud_Detection.git 
   

2. Install dependencies:

    python -m venv venv
    
    source venv/bin/activate  # For Linux/macOS
    
    venv\Scripts\activate.bat  # For Windows
    
    pip install -r requirements.txt
   

3. download data set from this Link Kaggle Dataset Link

4. edit path of data in code and run

Author

Muhammad Ruby