UK Regional Insights: A Geospatial Analysis of Economic GVA and Deprivation

This repository contains a data analytics project analyzing regional economic inequality in the United Kingdom. It implements an end-to-end ETL pipeline to process 26 years of geospatial and economic data from the Office for National Statistics (ONS).

The project then uses advanced geospatial statistics, time-series analysis, and interactive visualization to identify economic "hot spots," "cold spots," and long-term growth trends.

🚀 About the Project

The goal of this project is to explore the complex relationship between economic productivity (measured by Gross Value Added, GVA) and social-economic factors (like Population and the Index of Multiple Deprivation, IMD).

It moves beyond a simple static analysis by building a rich, time-series dataset spanning from 1998 to 2023. This allows us to ask deeper questions:

• Where are the statistically significant clusters of wealth and deprivation?
• Have these "hot spots" and "cold spots" changed over the last 26 years?
• What are the underlying growth trajectories of different regions?
• Which features (deprivation, location) are the best predictors of economic output?

✨ Packages

This project showcases a full-stack data science workflow, from data engineering to visualization.

---

🏗️ Project Structure

The repository is organized into a modular pipeline, separating data engineering from analysis and the final application.

UK-Regional-Insights/
├─ data_pipeline/
│  └─ transformers.py         # Python script to clean and merge all raw data
├─ notebooks/
│  └─ data_exploration.ipynb  # Jupyter notebook with all analyses
├─ app/
│  └─ streamlit_dashboard.py  # (Future) Code for the interactive web app
├─ models/
│  ├─ inequality_predictor.py     # (Future) Baseline ML models
│  └─ gnn_regional_networks.py    # (Future) GNN model
├─ assets/
│  ├─ animated-map-demo.gif   # Demo GIFs for this README
│  └─ lisa-hotspot-map.png    # Saved plots
├─ data/
│  ├─ input/       # (Ignored by git) Raw .xlsx and .geojson files
│  └─ processed/   # (Ignored by git) The final master GeoPackage
├─ outputs/        # (Ignored by git) Saved .html and .png visualizations
├─ .gitignore          # Ignores all data, output, and cache files
├─ environment.yml     # Reproducible Conda environment
└─ README.md           # You are here!

---

📊 Core Analyses & Visualizations

This notebook performs five key analyses to move from raw data to actionable insights.

1. Static & Animated Economic Maps

First, a 2D choropleth map (using folium) visualizes the GVA per capita for the most recent year (2023), clearly showing the static economic landscape. This is complemented by an animated time-series GIF (created from a folium.plugins.TimestampedGeoJson map) that shows the dramatic economic changes from 1998 to 2023.

2. Hot Spot & Cold Spot Analysis (LISA)

This analysis uses Local Moran's I (LISA) from the PySAL library to find statistically significant spatial clusters. It clearly identifies the "High-High" (Hot Spot) cluster around London and "Low-Low" (Cold Spot) clusters in former industrial areas and rural regions.

3. Economic "Winners & Losers" (CAGR Analysis)

This analysis moves beyond a static snapshot to identify long-term economic trajectories. It calculates the Compound Annual Growth Rate (CAGR) for GVA per capita for every Local Authority from 1998 to 2023. This metric reveals the "winners" (fastest-growing regions) and "losers" (stagnating or declining regions) over the past quarter-century.

4. Interactive Inequality Analysis (2D & 3D Scatter Plots)

These plotly charts visualize the complex, multi-dimensional relationship between GVA per Capita, Deprivation (IMD Rank), and Population. The 3D plot allows for a full exploration of how these three key variables interact, while linking the point's shape to the LISA clusters (Hot Spot/Cold Spot) connects this analysis back to the spatial data.

---

🔧 Getting Started

Prerequisites

This project uses Conda to manage its environment and dependencies. You'll need to have Anaconda or Miniconda installed.

Installation

1. Clone the repository:

   git clone [https://github.com/Tahernezhad/UK-Regional-Insights.git](https://github.com/Tahernezhad/UK-Regional-Insights.git)
   cd UK-Regional-Insights

2. Create the Conda environment: Use the provided environment.yml file to create the Conda environment. This will install all the necessary packages.

   conda env create -f environment.yml

3. Activate the environment:

   conda activate geoml

4. Download the Data: The raw data files are not included in this repository. Please download them from the official sources and place them in a data/input/ folder (you will need to create this folder).

   • GVA & Population: ONS Regional Accounts
   • IMD: English Indices of Deprivation 2019
   • Geometries: ONS Open Geography Portal

How to Run

1. Run the ETL Pipeline: Execute the transformers.py script to process all raw files into a single master GeoPackage.

   python data_pipeline/transformers.py

2. Run the Analysis Notebook: Launch Jupyter and open the main notebook to see all the analyses and visualizations.

   jupyter notebook notebooks/data_exploration.ipynb

🔮 Future Work

This project provides a robust foundation for predictive modeling. The next steps are:

• Streamlit Dashboard: Populate the app/streamlit_dashboard.py file to create a fully interactive web application.
• Baseline ML Model: Build a baseline RandomForest or XGBoost model to predict GVA_per_capita using the features from the exploration notebook.
• Graph Neural Network (GNN): Implement a GNN (using the models/ directory) to model the spatial network explicitly. The spatial weights matrix from the LISA analysis will serve as the graph's adjacency matrix, allowing the model to learn from neighboring regions. """