🌄 Setting the Dataset

After years of working in the banking industry, handling data in Excel, copying and pasting from reports, I realized that I wanted to truly understand the data and create actionable insights for the business: This project is my way out!!!

The aim is to understand the customers: what they do and what we could do to keep them!!

⚽ Main Goals

Set up the environment (PostgreSQL database with DBeaver)
Download and Work on the original banking dataset Ensure that the data follows the same distribution and patterns as the original dataset
Build a SQL schema to structure the data
Start analyzing customer segmentation and risk analysis

📋 Step by Step

📍 Step 1: Struggled with PostgreSQL setup in DBeaver (more on this later)
📍 Step 2: Expanded the dataset from 5K to 100K customers using Python and Jupyter Notebook
📍 Step 3: Applied SQL transformations to ensure consistency and structure
📍 Step 4: Ran the first queries to explore customer behavior and loan data

⚡ The Struggle with DBeaver & PostgreSQL

At first, I encountered a few headaches: configuring the PostgreSQL database with DBeaver wasn’t as smooth as expected. I spent hours troubleshooting connections, but after some trial and error, I finally managed to set it up.

Once the environment was working, I could finally focus on what mattered: getting data into PostgreSQL and starting the analysis.

📊 Dataset Expansion

I used the Banking-Dataset by Ahsan Habib as my base, a dataset containing 5,000 customers with various demographics, account details, loans, and credit card information. The challenge? Scaling it to 100,000 customers without losing the integrity of the original data.

I wrote a Python script in Jupyter Notebook to generate synthetic customer data, preserving the distribution and relationships in the original dataset. The final result was a 100K customer dataset ready for SQL analysis.

🔑 Key Insights

The expanded dataset allowed me to test a wider range of hypotheses, from customer segmentation to identifying risky loans.
The synthetic data generation process mirrored real-world banking dynamics while providing a robust dataset for deeper analysis.

⚙️ Code Snippet: Dataset Expansion with Python on Jupyter Notebook

import pandas as pd
import numpy as np
from datetime import datetime, timedelta
import random

original_data = pd.read_csv('banking_database.csv')

target_rows = 100000
existing_rows = len(original_data)
new_rows_needed = target_rows - existing_rows

print(f"Generating {new_rows_needed} new customer records...")

# Defining sample data to append on the existing dataset
first_names = ['Joshua', 'Mark', 'Joseph', 'Kevin', 'Linda', 'Charles', 'Christopher', 'Daniel', 'William', 

🎯 Next Steps

Run the first SQL queries on the expanded dataset, focusing on customer segmentation and loan analysis. Start building the Tableau dashboard to visualize and act on the data: understanding customers, spotting risks, and identifying upsell opportunities.