Who?

• Nik Lomax
  • Associate Professor of Data Analytics for Population Research
• Roger Beecham
  • Lecturer in Geographic Data Science, co-leader of the Centre for Spatial Analysis and Policy (CSAP)
• Layik Hama

What?

why (rationale)?

• We needed it even before the funding (2019)
• Geovisualization is tough
• Mainstream data science workflows has gaps for geovisualization
• Web enables interactivity in data science (both in R and Python)
• To discover new methods and techniques
• We can and are inspired by giants like Tableau et al.

“The most profound technologies are those that disappear”

~ (Weiser 1991)

So far

Paradigm

• RStudio and others drive JS from R (Shiny)
• TGVE keeps them (R & JS) separate completely

The npm package (1)

• Easy usage in various environments (web, mobile, desktop and GitHub pages)
  • data democratization
• API driven development
• Automating laborious tasks (Smart)
• Current challenge: modular and extensible

The npm package (2)

function App() {
  return (
    <Eatlas 
    defaultURL='https://api.coronavirus.data.gov.uk/v1/data?filters=areaType=utla&structure={"date":"date","newCases":"newCasesBySpecimenDate","ctyua19cd":"areaCode"}&format=csv'
    geographyURL='https://raw.githubusercontent.com/tgve/example-data/main/utlas.geojson'
    geographyColumn='ctyua19cd'
    column="newCases"
    dark="false"
    />
  );
}

R package: tgver (1)

# eval=FALSE
tgver::tgve()

knitr::include_url("https://tgve.github.io/app/?defaultURL=https://raw.githubusercontent.com/tgve/example-data/main/casualties_100.geojson")

tgver (2)

tgver (3)

• For (1) Advanced geospatial analysis, one vignette is in.
• for (2) more docs and emphasis? Shiny/IPyWidget comparison. Why not URL params?
• For (3) the package needs more functions. We can leave this for future versions

tgver (4) - time permitting

The R vignette: https://tgve.github.io/tgver/articles/r-and-js.html

Highlight (1)

API variables in URL

We can pass data and geography sources (all API values) to an instance of the TGVE.

• use the same instance with different datasets
• easy use in R and Python workflows (Rmarkdown & Jupyter respectively)
• like: www.instance.com?defaultURL=url&geographyURL=geoURL&geographycolumn=name
• future versions: www.instance.com?...filter={column:[1,2]}

Highlight (2)

Jupyter without IPyWidgets

https://www.kaggle.com/layikhama/jupyter-tgve

Showcase SPENSER

knitr::include_url("https://geospenser.com")

Showcase SaferActive

knitr::include_url("https://map.saferactive.org?")

Desktop

• Build Tools use Rust Lang
• Using WebAssembly & WebViews
• TGVE npm package in a CRA app is Tauri compliant
• More at: https://github.com/tgve/tgvejs

Build community

Future

• Obvious
  • Novel visualization layers
  • Automated geography detection
  • Dataset report generation
• AI module(s)?
  • VISAI?
  • Data profiling?

Conclusions

• Current functionality is enough to get the packages out
• Funding to collaborate with Turing REG (Research Engineering Group)
• Continue developing The TGVE
• Continue building a community around TGVE

Thank you

• ATI: www.turing.ac.uk/research/research-projects/turing-geovisualization-engine
• GH: www.github.com/tgve
• Recent visualization paper - (Beecham et al. 2021)

References