Pandas Vincent Vega D3 and Me and You and Your Cousin Too

Generating data visualizations programmatically is fun heya!

• It's a process:
  • Acquire - Parse - Filter - Mine - Represent - Refine - Interact
  • Ben Fry in Visualizing Data (O'Reilly)
• What can be done with some of these tools?
• Simple examples of Vega usage

---

Pandas

Open source library with data structures and data analysis tools for Python.

• DataFrame object for data manipulation
• Time series data functionality

DataFrame

A 2D labeled data structure with columns of potentially different types.

• Like a spreadsheet or SQL table, or a dict of Series objects
• Can pass in index and columns args (row and column labels)

Series

A one-dimensional labeled array capable of holding any data type.

• The axis labels together are called the index
• Will automatically missing data (NaN)

Vincent

Helps ease building of visualizations on top of D3 with Vega.

• Python data power, JavaScript visualization power
• Understands Pandas DataFrames and Series

Vega

A visualization grammar for creating and saving visualization designs.

• Declarative format
• Describe data visualizations in JSON
• Generate interactive views using HTML5 Canvas or SVG

Vega Runtime

Parses Vega specifications to produce interactive visualizations.

• With node.js can be run server-side to render specs to PNG or SVG
• Can also use the CLI tools vg2png and vg2svg
• Parser turns the JSON spec into a constructor for View objects (JS)
• Actual visualizations are instances of the vg.View class (JS)

Vega Specification

A Vega spec is simply JSON that describes the design of a Visualization

• The data sets used
• Scale transforms
• Encoding algorithms
• Axes
• Visual marks (e.g. lines, shapes)

• The specs are read and interpreted by the runtime to dynamically create viz (and more)

Vega Visualization

The top-level object in Vega - the container for all visual elements.

• Consists of:
  • A rectangular canvas, where the visual elements live
  • A viewport, a window to that canvas
• Basic properties, such as the view's width and height
• Definitions for:
  • The Data to visualize
  • Scales that map data values to visual values
  • Axes that visualize the scales (as ticks, labels)
  • Graphical Marks (shapes, arcs, etc) that visualize the data

D3

JavaScript library that provides visualization components and a data-driven approach for manipulating the DOM.

• Based on web standards (CSS3, HTML5, SVG)
• Focuses on the efficient manipulation of documents based on data
• Allows styles, attributes etc to be specified as functions of data
• Provides graphical primitives for charts (area, line, pie)
• Essentially a low-level library that can be used to build upon

Tools

Many good tools / libs / environments available.

• IPython: interactive shells; browser-based notebook
• NumPy: scientific computing with Python
• Anaconda: cross-platform distribution for data analysis etc
• Enthought: another cross-platform distribution; Canopy, EPD products