Task 1 Task 2 Task 3 Task 4 Task 5 Task 6 Task 7 Task 8 Task 9 Task 10

Portfolio Tasks

A collection of technical challenges completed throughout the Data Visualisation module, demonstrating skills in web development, data visualisation, APIs, and interactive design.

Portfolio Task 1

Hosting

Objective: Set up a GitHub account and live page using GitHub Pages. Display charts on your own site using the vegaEmbed function. Add two charts to the website: one chart from the course repository and one chart from an external web source. Ensure the live site is working correctly.

A Basic Chart

See below the added chart found in the repository.

A Chart from the Web

Another portfolio task is to add a chart from the web. Here is a chart that is not stored locally. This chart displays gun deaths and income. Using various US states, it shows median income on the x-axis and firearm deaths per 100,000 on the y-axis.

Portfolio Task 2

Building

Objective: Create your own visualisations using the Economics Observatory Data Hub. Build two separate charts using the "create" tool, export the JSON code and save to GitHub repository, then embed both charts in the webpage using vegaEmbed.

Adding in a Personal Chart

Portfolio Task 2 is to add our own charts. The following chart shows public sector employment in the UK.

Adding in Another Personal Chart

The second chart shows internal displacement in Bangladesh.

Portfolio Task 3

Debate

Objective: Use visualisation in policy commentary. Set out a policy topic (maximum 25 words), create two charts that support, refute, or relate to an argument on this topic, and add commentary (maximum 25 words) explaining findings.

Use Visualisation in Policy Commentary

Visualising the "aggregation problem" in global policy data

Should global progress on SDG 6 be judged by overall averages, or by improvements amongst the most disadvantaged groups?

The first chart shows that global access to safely managed drinking water has improved only marginally, suggesting progress towards SDG 6.1 may be minimal:

Disaggregated Analysis

Yet the second chart reveals faster improvement in rural areas, illustrating how disaggregated data exposes hidden progress and the need for targeted policy investment.

Portfolio Task 4

Replication

Objective: Re-create, then improve, someone else's chart. Find a chart from a policy organisation, journalist, or think tank, display the original chart as an image file, replicate the chart in Vega-Lite (matching axes, titles, colours), then improve the chart and explain changes (maximum 25 words).

Re-creating and Improving Someone Else's Chart

The following shows the Reuters "Gun deaths in Florida" chart. A reversed y-axis made it seem gun deaths declined after Florida's "Stand Your Ground" law in 2005, prompting backlash over how visual design can distort reality.

Reuters Gun Deaths in Florida Chart

My Replication

Below is my replication of the above chart.

Improved Version

And finally, here is an improved version of the chart. The revised graph corrects the inverted y-axis to ensure accurate interpretation and incorporates interactive tooltips to improve data transparency and user comprehension.

Portfolio Task 5

Accessing Data

Objective: Access data through APIs and web scraping. For the API task, add a chart using a live API link and include a functional description of the API (base URL, elements, final URL). For the scraper task, scrape a website using Google Colab, clean and normalise data, export in tidy format, create a chart from scraped data, and comment on the data source and challenges (maximum 25 words).

API Task

The following shows data on crime in London, accessed via an API.

API Functional Description

Base URL: https://data.police.uk/api/

Endpoint: crimes-street/all-crime

Parameters:

  • lat - Latitude coordinate (e.g., 51.5074 for central London)
  • lng - Longitude coordinate (e.g., -0.1278 for central London)

Full URL used: https://data.police.uk/api/crimes-street/all-crime?lat=51.5074&lng=-0.1278

Scraper Task

Data collection and analysis using web scraping techniques. The chart below visualises data extracted through web scraping and processed in Python.

I explored the frequency of certain words used on news websites when reporting crime in their headlines. I compared BBC News, The Guardian, Sky News and GB News.

View the full analysis:

Scraper Task Notebook
Portfolio Task 6

Loops

Objective: Build a dashboard using loops for batch downloading and chart embedding. Use a loop to batch download six different time series as JSON files, save files to GitHub and use raw files to supply data, display six charts in a grid layout (3×2 or 2×3), use a JavaScript loop to embed all six charts programmatically, and link to Google Colab notebook showing the data download process.

Building a Dashboard

The following shows a dashboard built using JavaScript loops and multiple food Producer Price Index datasets, demonstrating programmatic chart generation.

View the Python notebook showing the data download process using loops:

Dashboard Notebook
Portfolio Task 7

Maps

Objective: Create base maps and choropleths. Produce two maps: one of Scotland and one of Wales. One map should be a coordinates map (point data) and one map should be a choropleth (shaded regions). Choose any data of interest to visualise.

Choropleth Map (Scotland)

The following shows a choropleth map of happiness scores in Scotland by Local Authority Districts, demonstrating geographic data visualisation.

Coordinates Map (Wales)

Here is a coordinates map of main GP Sites in Wales, plotting point data on a geographic canvas.

Portfolio Task 8

Big Data

Objective: Extract a story from millions of prices using UK retail pricing databases. Produce two charts using UK prices datasets (LRPD or AutoCPI), simplify or reduce the data before plotting (grouping, aggregation), and explain what you have done (maximum 50 words). Avoid plotting raw scatter data with thousands of observations.

Data Processing Methodology

Both charts use the Long Run Prices Database (LRPD), containing millions of price observations from 1988 onwards. To extract meaningful insights, I grouped data by product category and time period, calculating average prices to reduce dimensionality. The cake analysis aggregates prices monthly to show inflation trends, whilst the tampon analysis groups by region to reveal geographic pricing disparities across the UK.

Making a Cake Over The Years

Analysis of cake pricing trends over time using large-scale retail pricing data.

View the process:

Average Cake Price Notebook

Average Tampon Prices by Regions

Regional price analysis for feminine hygiene products, exploring geographic pricing patterns.

View the process:

Average Tampon Price Notebook
Portfolio Task 9

Interactive Charts

Objective: Produce two charts with meaningful interactivity. Create two interactive charts using advanced interactivity such as sliders, drop-down boxes, or clickable legends. Simple tooltips and colour selection do not count. Interactivity should help users better understand the data.

Paracetamol Prices Over The Years

Interactive analysis of painkiller pricing trends with dynamic filtering and tooltips.

Suicide Rates by Social Group in England

Analysis of mental health outcomes across different social demographics, with interactive exploration.

Portfolio Task 10

Advanced Analysis and Machine Learning

Objective: Conduct advanced analytics and machine learning analysis. For advanced analytics, create a chart using techniques beyond standard line, bar, or scatter charts (such as bubble charts, histograms, de-trended analysis, diff-in-diff, or heat maps). For machine learning, apply supervised (regression or classification) or unsupervised (clustering) learning. Transform data into standardised X matrix and y vector format, use sklearn in Python, state hypothesis (maximum 25 words) and create at least one visualisation, then comment on findings (maximum 25 words).

Part 1: Comparing Weight Issues in England

Statistical analysis of health data across English regions using advanced analytical techniques.

Part 2: Life Expectancy Regression Analysis

Hypothesis

Higher income deprivation scores are associated with lower life expectancy, with this relationship being stronger for males than females across English local authorities.

Machine learning regression model predicting life expectancy based on socioeconomic and health indicators. This analysis explores the relationships between various factors and population health outcomes.

The interactive visualisation below shows the regression results with confidence intervals:

Regression Results

Male Model:

R² Score: 0.7188

RMSE: 0.6927 years

Coefficient: -1.1075

Intercept: 79.7256

Female Model:

R² Score: 0.4553

RMSE: 0.9238 years

Coefficient: -0.8446

Intercept: 84.0609

Key Findings

The hypothesis is confirmed: income deprivation significantly predicts life expectancy. The male model shows substantially stronger predictive power, suggesting socioeconomic factors affect male mortality more acutely than female.

For detailed methodology, model training, and statistical analysis, view the complete Python notebook:

Regression Analysis Notebook