https://github.com/sinclairn101/dbt-nz-banking-loans

A personal dbt project transforming a synthetic New Zealand banking loans dataset into a star schema with staging, dimension, fact, and reporting models.
https://github.com/sinclairn101/dbt-nz-banking-loans

analytics business-intelligence dbt portfolio-project postgres sql

Last synced: about 1 month ago
JSON representation

A personal dbt project transforming a synthetic New Zealand banking loans dataset into a star schema with staging, dimension, fact, and reporting models.

• Host: GitHub
• URL: https://github.com/sinclairn101/dbt-nz-banking-loans
• Owner: SinclairN101
• Created: 2025-09-21T22:07:03.000Z (9 months ago)
• Default Branch: main
• Last Pushed: 2025-09-21T22:23:55.000Z (9 months ago)
• Last Synced: 2025-09-21T23:31:25.074Z (9 months ago)
• Topics: analytics, business-intelligence, dbt, portfolio-project, postgres, sql
• Language: YAML
• Homepage:
• Size: 1.58 MB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
# 🏦 NZ Banking Loan & Fraud Risk — dbt Project

This is a portfolio project where I practice **SQL** and **dbt** using a dataset in the lending space.  

The raw data comes from Kaggle: [**Synthetic NZ Banking Loan and Fraud Risk Dataset**](https://www.kaggle.com/datasets/sandycandy7/synthetic-nz-banking-loan-and-fraud-risk-dataset).  

I transformed the dataset into a layered analytics warehouse with clearly defined staging, dimension, fact, and reporting models. The goal was to replicate how real-world analytics teams structure data for reliable analysis and decision-making.

 ### 🎯 Goals

- Build a dbt project starting from raw seeded data  

- Design and implement a simple star schema  

- Write clear SQL transformations to create clean, analysis-ready tables  

- Apply dbt tests and documentation to ensure quality and transparency  

- Strengthen my end-to-end data & analytics skills

### 🧱 Project Structure

- **Seed (Raw Data)**  

  The Kaggle CSV is loaded into the warehouse with `dbt seed`. This acts as the upstream input for staging.

- **Staging (`stg_`)**  

  Cleans and standardizes the seeded table:

  - Renames columns to clear, readable names  

  - Casts 0/1 flags into true/false booleans  

  - Normalizes categories (e.g. `application_type`)  

  - Converts interest rates into percentages with proper numeric casting  

- **Dimension (`dim_`)**  

  Descriptive fields you group by or filter on.  

  Example: `dim_loan_applications` holds borrower demographics, credit grade, loan purpose, and dates.

- **Fact (`fct_`)**  

  Numeric outcomes and risk metrics you aggregate.  

  Example: `fct_loan_performance` stores loan amounts, credit scores, utilization ratios, inquiries, and fraud/default flags.

- **Reporting (`rpt_`)**  

  Pre-aggregated metrics that answer business questions.  

  Example: `rpt_loan_summary` shows total loans, averages, default counts, and default rates.

### 📊 Example Metrics from `rpt_loan_summary`

- Total loans: ~5,000  

- Total loan amount: ~95M  

- Average loan amount: ~18.9K  

- Average interest rate: ~19.9%  

- Average debt-to-income ratio: ~0.18  

- Average funding days: ~7.5  

- Total defaults: 48  

- Default rate: ~0.96%  

### 📚 Documentation and Testing

- Each model has a YAML file with column-level descriptions.  

- Documentation is written in `docs.md` blocks for clear, human-readable explanations.  

- Tests include `unique` and `not_null` checks on keys like `loan_id`.  

- I used `dbt docs generate` and `dbt docs serve` to build an interactive documentation site with lineage graphs and column-level details.  



  

  



### 🔍 Analysis & Dashboard

To showcase the value of the transformed data, I built a Metabase dashboard directly on top of the dbt models.  

This dashboard provides both a high-level snapshot of the loan portfolio and deeper risk insights.

### 📊 Portfolio Snapshot

The top row shows key metrics at a glance:  

- **Total Loans**  

- **Total Loan Amount**  

- **Average Loan Size**  

- **Average Credit Score**  

- **Default Rate (%)**  

- **Average Funding Days**

These KPIs summarize the scale and health of the lending portfolio.

### 🧩 Portfolio Mix

- **Loans by Purpose** 

- **Loans by Region** 

- **Loans by Credit Grade** 

### ⚖️ Risk & Performance

- **Defaults by Credit Grade**  

- **Defaults Over Time vs Total Loans**

- **Average Interest Rate by Grade**

### 🚀 Operational Insights

- **Funding Speed Distribution**  

- **Top 10 Borrowers by Loan Amount**

### 📸 Dashboard