Data Visualization

  • Edward Tufte’s graphical integrity
    • Show the Data
    • Maximize Data-Ink Ratio
    • Erase Non-Data Ink
    • Multi-functioning Graphical Elements
    • Data Density and Small Multiples
    • Use Clear, Detailed Labels
  • Beware of axis manipulation

Libraries

  • Python static plotting
  • Pandas uses matplotlib as backend by default, but it’s possible to switch to Plotly.
  • Plotly, based on plotly.js, which is again based on D3.
  • Bokeh, primarily Python, but also provides bokeh.js, uses Tornado to serve.
  • Altair, produces chart spec in JSON, rendered in browser using Vega-Lite, abuilt on top of Vega and D3.