Vega-Lite implementation of https://demo.baremetrics.com/forecast chart.
git clone https://github.com/alexander-matsievsky/vega-lite-evaluation.git /tmp/vega-lite-evaluation
cd /tmp/vega-lite-evaluation
python -m SimpleHTTPServer 8080 # python2
python -m http.server 8080 # python3
browse http://localhost:8080Since datavis lies at the heart of https://demo.baremetrics.com app streamlining charts' production is highly desirable for rapid development and ease of maintenance.
Vega-Lite is a high-level grammar of interactive graphics. It provides a concise JSON syntax for rapidly generating visualizations to support analysis.
It would be nice to have a library of plug-n-play declarative visualization specifications backed by a trusted platform and ecosystem.
Implementing a static version of the chart was enjoyable. D3 skills felt right at home. Overall Vega-Lite seems to be very well thought through and the enthusiasm of the community is encouraging. There are several quite intriguing dataviz solutions supporting Vega, e.g. Altair or Observable to name some of my favorites.
Nonetheless the current state of Vega-Lite is not well suited for custom interactivity. A more low-level Vega grammar supporting Signals must be employed leading to prohibitive development and maintenance costs.
Vega project is obviously a thing to keep a close eye on. Right now I'd pick it for quick throw-away visualizations during exploration or in cases where granular interactivity is not needed. For a customized interactivity of baremetrics.com good ol' D3 wrapped into Vue components library would serve.
P.S.: the repo contains the static version of the chart only. I tried to go for the Vega implementation but quickly realized that the output would be cryptic and horrifying+)