Python package for the WAYS project. See API documentation.
“As you can see in figure 1…” may well be the most frequently made claim in science. But unlike claims concerning data, statistics, models and algorithms, those relating to visualisations are rarely evaluated or verified. So how can data scientists understand visualisations’ effectiveness and expressiveness? What is the visualisation equivalent of q-q plots, R^2 and K-folds tests?
Designing effective visualisations goes far beyond selecting a graph, scales and a ‘pretty’ style. Effective visualisations must negotiate sensitivities and interactions between visual elements (e.g. encodings, coordinate systems, guides, annotations), data (e.g. characteristics, transformations, partitions), and the discriminator function, which in this case is the perceptual and cognitive systems of humans. Despite their criticality, these methodological and design considerations are rarely surfaced, limiting the value extracted from visualisations. What does figure 1 actually visualise?
The ‘What Aren’t You Seeing’ (WAYS) project addresses 1) what we aren’t seeing in visualisations by 2) revealing the relevant knowledge, theory and practices that we are not seeing at the site of visualisation production. Our final goal is the WAYS package/library in which the properties, outcomes and affordances of visualisation designs are depicted through visualisations; a concept we term ‘Precursor Visualisations’. WAYS then addresses the challenge of generating a productive interplay between everyday visualisation work and the epistemology, practice, communication techniques and evaluation methods that should inform visualisation design at source (Robinson). To achieve this, we propose three work packages (WP1-3).
Install from PyPI using pip install ways-py.
To create the development environment:
- Install python poetry by following the install instructions
- From the top level dir, run
poetry shell
Your machine will need Python 3.9 ) available.
After creating the above development environment, clone this repo and navigate into it:
git clone https://github.com/WarwickCIM/ways-py.git
cd ways-pyThen run Jupyter like so:
poetry run jupyter notebookNavigate to the main demo notebook at notebooks/usa-presidential-poll.ipynb.
We use a number of git aliases to track different kinds of commit; to use these yourself, add an [include] section to your .gitconfig.