Bank Scoring Dashboard Project

Welcome to the Bank Scoring Dashboard project repository! This project offers an innovative credit scoring tool designed to calculate the probability of a client repaying their loan, as well as an interactive dashboard aimed at enhancing customer relationships.

Overview

The main goal of this project is to implement a credit scoring tool to evaluate the likelihood of a client repaying their loan. Additionally, we aim to develop an interactive dashboard dedicated to customer relationship management, providing bank advisors with valuable insights into clients' financial situations.

Key Features

• Credit Scoring Tool: This tool calculates the probability of a client repaying their loan based on their financial data.
• Interactive Dashboard: An intuitive dashboard that allows bank advisors to visualize clients' financial data, thereby enhancing customer relationships.
• Model Interpretability: The project ensures that the model's predictions are understandable and easy to communicate, assisting bank advisors in their decision-making process.

Challenges

The project addresses the following challenges:

• Cleaning and extracting important information from multiple datasets
• Training and optimizing a classification model with imbalanced datasets
• Extracting useful data for the development of an interactive customer relationship dashboard
• Deploying the dashboard online via a VPS

Approach

The project follows a three-step approach:

1. Data Import and EDA (Exploratory Data Analysis) for each dataset:

• Handling missing values
• Analyzing correlations
• Performing feature engineering

2. Training and Evaluation of the LGBMClassifier model:

• Simple training
• OverSampling
• UnderSampling
• Custom metrics with threshold

3. Model Interpretability:

• Feature importance
• Selection of important variables for the dashboard

Project Structure

The project is organized into two main parts: the scoring model and the interactive dashboard. The project files are as follows:

1. Scoring Model Code

• functions.py: Main functions used in the notebook
• notebook_scoring.ipynb: A notebook for exploration and modeling, including preprocessing steps adapted from an existing Kaggle kernel, which can be found here

2. Interactive Dashboard

• dashboard_app_streamlit.py: The code for the dashboard, developed with Streamlit
• notebook_prep_API.ipynb: A file containing tests for the application and the creation of additional tables

Getting Started

To begin using this project, follow these steps:

1. Clone the repository: git clone https://github.com/yourusername/bank-scoring-dashboard.git
2. Navigate to the project directory: cd bank-scoring-dashboard
3. Install the required dependencies: pip install -r requirements.txt 4 Run the Streamlit dashboard: streamlit run dashboard_app_streamlit.py

By following the above steps, you will have access to the interactive dashboard, which serves as a valuable tool for bank advisors. The dashboard is packed with various graphs and metrics that allow users to effectively evaluate a client's financial situation.

Feel free to explore the code, contribute to the project, and share your insights!

---

This project is a comprehensive and user-friendly solution for bank advisors looking to enhance their customer relationships and make informed decisions regarding loan approvals. With an easy-to-understand credit scoring tool and an interactive dashboard, bank advisors can efficiently evaluate clients' financial situations and provide valuable insights to their clients.