This project aims to develop a comprehensive risk management API for the Bank of Baroda. The API includes several features such as credit card fraud detection, loan fraud detection, a 24/7 helping chatbot named BOB, user activity tracking, and a personalized recommendation system for investments. The main codebase is developed using Jupyter Notebooks and utilizes various Python libraries for data processing, balancing, and modeling.
- Credit Card Fraud Detection: Identifies fraudulent credit card transactions.
- Loan Fraud Detection: Detects potential fraudulent loan applications.
- 24/7 Helping Chatbot (BOB): Provides round-the-clock assistance to users.
- User Tracking: Monitors user activity within the application.
- Personalized Recommendation System: Suggests investment opportunities and plans based on user preferences and goals.
pandas: Data manipulation and analysis.numpy: Numerical operations.matplotlib: Plotting and visualization.seaborn: Statistical data visualization.
imblearn: Handling imbalanced datasets.SMOTE: Synthetic Minority Over-sampling Technique.RandomUnderSampler: Random under-sampling of the majority class.Pipeline: Utility for creating a composite estimator.
sklearn: Machine learning library.train_test_split: Splitting datasets into training and testing sets.confusion_matrix: Evaluating the accuracy of classification.roc_auc_score: Calculating the area under the ROC curve.RocCurveDisplay: Visualizing the ROC curve.cross_val_score: Cross-validation of the model.GridSearchCV: Hyperparameter tuning.classification_report: Summarizing the performance of the model.RepeatedStratifiedKFold: Cross-validator with stratified folds.precision_recall_curve: Plotting precision-recall pairs.roc_curve,auc: Calculating and plotting ROC curves.
The project uses various datasets for training and testing the models. The datasets are preprocessed and balanced to ensure accurate and reliable results.
bash git clone https://github.com/your-username/risk-management-api.git cd risk-management-api
'bash jupyter notebook Open the main_notebook.ipynb file to start working with the code.
- Credit Card Fraud Detection
- Load the dataset and preprocess it.
- Use SMOTE and RandomUnderSampler for data balancing.
- Train the model using the balanced dataset.
- Evaluate the model using ROC, precision-recall curves, and other metrics.
- Similar steps as credit card fraud detection, with a different dataset.
- 24/7 Helping Chatbot (BOB)
- Implement the chatbot using a suitable framework (e.g., Rasa, Dialogflow).
- Integrate the chatbot with the API.
- Monitor user activity and store the data for analysis.
- Use user data to provide personalized investment recommendations.
- Implement algorithms to suggest investment plans based on user goals.
python start
--Runs the app in development mode.
python test
--Launches the test runner.
python run build
--Builds the app for production.
python run eject
--Ejects the project from its current configuration.\