Global Wealth and Human Development Visualisation

An interactive D3.js visualisation developed for MSc Artificial Intelligence Data Visualisation coursework at King's College London. It explores country-level wealth, education, and human-development patterns from 2000 to 2020 through linked geographic, statistical, and temporal views.

What the application shows

The left panel is a choropleth of GDP per capita. The right panel plots GDP per capita against secondary-school enrolment, with colour identifying World Bank regions.

The implemented interactions are:

• a year slider that updates both views from 2000 through 2020;
• country hover tooltips in both views for GDP per capita, enrolment, HDI, life expectancy, and region;
• country selection from either view, linking the views by highlighting the map and drawing that country's scatter-plot trajectory;
• regional filter buttons for isolating or comparing geographic groups; and
• animated transitions as the selected year changes.

Visual design

• Geographic view: an equal-area Eckert VI projection preserves area comparison better than a Mercator projection. GDP per capita uses a sequential YlOrRd scale because the variable is ordered.
• Colour scaling: the colour domain is capped at the 97th percentile and transformed with a 0.4 power exponent. This reduces domination by very high-income outliers while retaining contrast among lower-income countries.
• Statistical view: a linked scatter plot uses linear axes for GDP per capita and gross secondary-school enrolment. Region is encoded with a categorical colour scale.
• Time: the shared year slider provides an annual snapshot, while country focus mode adds a 2000-2020 trajectory for longitudinal inspection.

The visualisation is exploratory. It shows associations and geographic patterns; it does not establish causal relationships.

Data preparation

The implemented preprocessing pipeline reads two local coursework inputs:

1. Global_Development_Indicators_2000_2020.csv supplies country, region, GDP per capita, HDI, life expectancy, population, and fallback enrolment fields.
2. world-education-data.csv supplies the preferred secondary-school enrolment field.

The script filters to ISO3 country codes and 2000-2020, left-joins the enrolment data by country and year, records which source supplied enrolment, maps ISO3 codes to the numeric identifiers used by the map geometry, and writes a nested country/year dataset.

The committed derived dataset contains 217 country objects and 4,548 country-year records. Missing numeric observations are written as JSON null values, and strict serialisation rejects NaN or Infinity.

Raw coursework and module-provided CSV files are not redistributed here. The separate mean-years-of-schooling-long-run.csv file was used for analytical comparison in the report but is not read by the implemented web-application pipeline.

Run locally

The application must be served over HTTP so the browser can fetch the JSON files:

python -m http.server 8000

Then open http://localhost:8000 in a browser. The page loads D3, TopoJSON, and the Eckert VI projection helper from public CDNs, so an internet connection is required for those libraries.

Rebuild the derived dataset

Obtain the two raw inputs under their original access and redistribution terms, then keep them outside the repository.

python -m venv .venv
python -m pip install -r requirements.txt
python preprocess.py --source-dir "path/to/coursework/Software"
python scripts/validate_data.py

The source directory must contain:

• Global_Development_Indicators_2000_2020.csv
• world-education-data.csv

By default, preprocessing writes data/countries.json. Use --output to select another derived-output path.

Validation

python scripts/validate_data.py

The validator strictly parses every committed JSON file, rejects non-standard numeric tokens, checks the country/year schema, and confirms the expected country and record counts.

Data sources and attribution

• The primary coursework input is Global Development Analysis (2000-2020) by Michael Matta on Kaggle. Its data card describes a dataset curated from public data through Google BigQuery; the coursework report identifies World Bank development indicators as the principal underlying source.
• The World Bank World Development Indicators provide the underlying economic and development context described in the report.
• Map geometry is from TopoJSON World Atlas, derived from Natural Earth country boundaries.
• Our World in Data's average-years-of-schooling series, based on Barro-Lee and Lee-Lee research, informed report-level comparison but is not part of the implemented preprocessing pipeline.

The exact upstream source and redistribution terms for the coursework-supplied world-education-data.csv file have not yet been independently verified. That raw file is therefore not included. The committed countries.json is a derived visualisation dataset; its continued public redistribution should be reviewed alongside the source terms before a licence is chosen.

Limitations

• Coverage is restricted to 2000-2020 and to records that can be mapped through the project's ISO code lookup.
• Secondary-school enrolment remains missing for 1,468 of 4,548 country-year records after the two input sources are combined.
• Gross enrolment can exceed 100 percent because it may include over-age and under-age students.
• GDP per capita is a national average and does not represent within-country inequality.
• Mixed source coverage and annual resolution limit direct comparability across every country and year.
• The views support exploratory association and comparison, not causal or predictive conclusions.

Repository structure

index.html                 Linked D3 visualisation
preprocess.py              CSV joining and strict derived-JSON generation
data/countries.json        Committed derived country/year data
data/world-110m.json       TopoJSON map geometry
scripts/validate_data.py   Strict JSON and schema validation
assets/                    Public-safe application screenshots

No licence has been selected for this repository.