Find answers, ask questions, and connect with our
community around the world.

Activity Forums Web Design Best libraries/techniques for large-scale data visualization?

  • Best libraries/techniques for large-scale data visualization?

    updated 3 weeks ago 0 Member · 1 Post
  • Maverick

    Member
    October 29, 2019 at 12:47 pm

    My master thesis consists on a tool that allows both experienced and amateur astronomers users to better exploit the ALMA Science Archive through data mining and visualization – this question will be focused on the latter. First of all, I don’t have that much of a background in web development, so these past few weeks have involved a lot of trial-and-error experimenting with different data visualization Javascript libraries. D3 does seem to be the obvious choice but, once I tried to implement a very simple scatterplot showing the coordinates of every single observation taken by ALMA (which currently amounts to about 35,000 data points), I quickly realized that that library alone doesn’t perform quite well with large datasets – which seems to be a very consensual take. I looked around for alternatives and found these. Also, keep in mind most of my remarks on each are more of a matter of personal perception; I’d love to have my opinions on each thorougly challenged 🙂 Plotly.js has a gentle learning curve and more of a “high-level” feel to it. Great for plotting quick charts, but doesn’t seem as customizable as vanilla D3. The built-in support for WebGL rendering is nothing to look over! D3 + Canvas seems to tackle the performance issues quite well; however, when it comes to interactivity and animations things start to get finicky as it requires a lot of working-around – which could unnecessarily complicate the code. I’ve had some moderate success converting SVG visualizations to Canvas, but the performance was still subpar. Stardust is a strong contender, but I wasn’t able to find a full tutorial on how to implement zoom/panning/tooltips and other gizmos that could come in handy. Apparently, those can be managed by D3 – alongside with the axis – but the whole deal sounds kinda confusing. Now, I can hear the collective facepalming even before posting this – I’m 99% certain my conclusions and opinions on this problem are inaccurate at best. As a near-complete newbie when it comes to data visualization, sites running as smoothly as (for instance) Windy does seem like dark magic, and I’d very much like to know this power. TL;DR: what libraries and/or techniques should be used for web-based, large scale data visualization? I’m prioritizing interactivity and performance, which almost goes without saying. – by hq overview TheGreatDaiamid – –

Reply to: Maverick
Your information:

Cancel
Original Post
0 of 0 posts June 2018
Now