Links
- public github repository
- contribution to the Scottish Tech Army Climate Dashboard – code for front-end, visual analysis of active travel across Scotland
- some publications documenting previous work – on the dblp computer science bibliography
- supplementary information for publications on figshare
Visual Storytelling
Unfortunately a good part of my work sits behind IP and commercially sensitive walls. Luckily, I’ve also worked on some projects that required or at least encouraged openly sharing results, typically with a CC BY license. The latter typically allow more room for innovation and exploration, and the skill I developed and knowledge I gained on those have ultimately fed back into the work that sits behind those walls! A selection of the open interactive visualisations I’ve built —
as demonstrators to illustrate the outcomes of work to:
- model, merge and/or build on open datasets, to provide a consistent, single source of truth for wider reuse
- perform root cause analysis during issue resolution
to uncover and tell stories contained within data, to:
- guide context-sensitive analysis and “follow your nose” to uncover relationships within highly inter-linked data
- track and explore evolving trends, as for pupil and teacher demographics in Scotland, working with dressCode and the Scottish Tech Army
- provide insight into complex, multi-source, high-dimensional and/or inter-related data, as seen in a summary of changes to active travel data across Scotland prior to and during the 2020 and 2021 lockdowns. (See also the interactive version)
- a related visualisation looks at the biannual all-modes transportation survey carried out at selected locations in each of Scotland’s 32 local authorities. (r code available on request)
- serve as an agile, adaptable, working tool to explore alternative scenarios, as when modelling operations within a digital business ecosystem
Modelling & Mastering Data
Invariably, building each visualisation meant reusing domain and/or third party (open) data. Where exploring new scenarios this sometimes meant initially creating dummy data on top of the data model built with domain owners and/or end users. For third party data sometimes this meant deriving the data model first, in order to extend this and/or build extensible visualisations directly on top of the data and associated situation models.
I am a visual thinker, so learning early in my career to work with ontologies and graph models, predominantly on semantic web data-driven projects, has been a blessing in more ways than I can count. Especially since translating these models into simple node-link graphs that mimic mind maps allows pretty much anyone to follow relationships within the data, and quickly obtain an overview of a dataset, without needing a background in complex data analysis or graph models.
Experience has also taught me modelling data and building or cleaning with an eye to reuse means that over time you recoup the up to 80% of project time that may be required to clean and preprocess data – with reusable data pipelines, cleaning code and even mastered data subsets in other data projects.
A couple examples of data models I’ve built, to feed into visualisation-guided analysis:
- Energy from Waste ontology, to feed into modelling operations and quantifying value within a digital business ecosystem
- Scottish Qualifications Authority data – pupil grades & teacher demographics
- Skill and job role similarity analysis in an EU-funded project investigating human resource and skills in data science, to satisfy growing demand for the data-driven economy
Situation & scenario modelling
- High level summary of proposal for a green-themed business presented to the judges, that won me a scholarship to study for an executive MBA