🚀 End-to-End Banking Analytics & Predictive Risk Platform

An enterprise-grade banking data intelligence project that leverages relational databases, statistical analysis, and machine learning to optimize customer retention, flag fraudulent transactions, and mitigate credit portfolio risk.

---

🎯 Project Overview & Core Goals

Modern financial institutions process massive transactional streams daily. This multi-layered analytical engine transforms raw relational database structures into actionable business frameworks.

The primary business objectives achieved are:

1. Customer Segmentation & Retention: Tracking behavioral decay to predict user churn.
2. Fraud Detection & Auditing: Isolating anomalous spending behaviors in real time.
3. Credit Risk Management: Quantifying structural capital default exposure.
4. Revenue Distribution Mapping: Visualizing transactional volume contributions across premium card tiers.

---

🧰 Tech Stack & Architecture

• Language: Python 3.10+
• Core Libraries: Pandas, NumPy, Scikit-Learn
• Visualizations: Matplotlib, Seaborn
• Database Training: Relational SQL Query Workflows
• Application Deployment: Streamlit Web Framework

🏗️ Relational System Blueprint

The platform operates on a robust, three-table relational database architecture joined together by the CustomerID primary key string:

   [ Customer Profiles ]
         (CustomerID)
              |
      +-------+-------+

      |               |
      v               v
[ Loan Records ]  [ Transaction History ]

---

📁 Repository Directory Structure

Organize your project workspace folder using this standardized directory mapping layout:

├── data/
│   ├── customer_profiles.csv     # Demographic attributes & Credit tiers
│   ├── transaction_history.csv    # 1,500 continuous chronological streams
│   └── loan_records.csv          # Active institutional debt liabilities
├── scripts/
│   ├── data_generator.py         # Relational database synthesizer (Block 1)
│   ├── churn_predictor.py        # Scikit-Learn training script (Block 2)
│   └── dashboard_plot.py         # Power BI visual matrix output engine (Block 3)
├── database/
│   └── banking_queries.sql       # Advanced CTE/Windowing script (Block 4)
├── streamlit_app.py              # Interactive Front-End user portal (Block 5)
├── banking_powerbi_dashboard.png # High-Resolution Executive Dashboard image
└── README.md                     # Comprehensive project portfolio manual

---

🧠 Core Features & Implementations

1. Data Synthesizer Engine

• Simulates realistic customer age arrays and income metrics.
• Models everyday consumer habits using an exponential distribution function curve.
• Injects mathematical fraud spikes (1.5% frequency) mimicking high-magnitude capital theft events ($3,000–$12,000 outliers).

2. Predictive Churn Classification (Scikit-Learn)

• Engineers behavioral features like Recency (days since active transaction) and Transaction Counts.
• Utilizes a Random Forest Classifier to assess churn risks.
• Addresses low-density class imbalance through a targeted, in-memory synthetic minority oversampling strategy.

3. Power BI-Style Executive Visualization Grid

• Combines four comprehensive analytical reports inside a unified, widescreen workspace canvas plot.
• Panel A: Machine Learning Feature Importance Drivers bar chart.
• Panel B: Asset Delinquency default rate metrics per credit group.
• Panel C: Transaction Auditing outlier scatter plot tracking standard deviation barriers ($Z$-Score $> 3$).
• Panel D: Revenue share breakdown pie visual per account category.

---

💡 Key Analytical Insights Documented

• Behavioral Churn Weights: Customer Recency (55.6%) and Transaction Frequency (26.0%) dominate the machine learning model's predictive importance. Customers consistently reduce their transaction velocity long before closing an account.
• Risk Profiles: Loan accounts associated with a Poor Credit Classification (<580) carry significantly higher default probabilities than Excellent Tiers (740+), confirming a clear statistical link between credit ratings and debt risk.

---

🚀 How to Set Up and Run the Platform

Follow these steps to deploy the complete environment pipeline on your local computer:

1. Clone the Repository Workspace

git clone https://github.com
cd banking-analytics-platform

2. Install Required Package Dependencies

pip install pandas numpy matplotlib seaborn scikit-learn streamlit

3. Run the Sequential Pipeline

# Step A: Instantiate the relational database files
python scripts/data_generator.py

# Step B: Execute the Scikit-Learn machine learning model code
python scripts/churn_predictor.py

# Step C: Save out the High-Resolution dashboard matrix image
python scripts/dashboard_plot.py

4. Boot Up the Interactive Streamlit Web Interface

streamlit run streamlit_app.py

---

📊 Sample Executive Dashboard Output

Upon running the dashboard script, a publication-grade graphics layout file titled banking_powerbi_dashboard.png will be generated inside your workspace directory, displaying the 4-panel business grid.

---

💡 Project developed for professional portfolio assessment and advanced data engineering training via CareerDream.in.