Data Archive

Altair

What Is Altair?

Altair provides a declarative interface where you specify what you want to visualize (e.g. variables, encodings, transformations), and the library handles how to render it.

It is powered by Vega-Lite, a JavaScript-based visualization grammar, and outputs charts as HTML/JSON that can be embedded in Jupyter notebooks, web apps, and dashboards.

Core principles:

  • Declarative syntax: You describe relationships between data fields and visual properties (e.g. map x to time, color to category).
  • Interactive by default: Charts can support hover, zoom, selection, etc.
  • Data-aware: Built-in data transformations like aggregation, filtering, binning.
  • Consistent and safe: Prevents many bad visualization practices by design (e.g., overplotting, misaligned axes).

Why Use Altair?

1. Readable and Concise Code

Altair lets you build complex visuals with minimal syntax.

import altair as alt
import pandas as pd
 
df = pd.read_csv('data.csv')
 
alt.Chart(df).mark_bar().encode(
    x='category:N',
    y='sales:Q',
    color='region:N'
)

You don’t need to specify chart layout, axes, or legends — Altair infers them.

2. Interactivity Built-in

With .interactive(), you can pan, zoom, and hover out-of-the-box — no extra code.

alt.Chart(df).mark_circle().encode(
    x='x_val', y='y_val'
).interactive()

You can also create complex interactivity like brushing, linking, or filtering across charts.

3. Data Transformation Within the Chart

Altair supports:

  • Filtering
  • Aggregating
  • Window functions
  • Binning
  • Calculated fields

You can express a lot of preprocessing without leaving the plotting context.

4. Best for:

  • Exploratory data analysis (EDA)
  • Interactive data dashboards (e.g. with Streamlit, Voila, or Panel)
  • Communicating insights with clean, web-ready charts

When Not to Use Altair

  • If you need fine-grained control over static plots for publication.
  • If you’re working with large datasets (>5000 rows by default; though workarounds exist).
  • If you’re deeply invested in Matplotlib or want full control over low-level rendering.

Example Use-Cases

  • Time series dashboards: Interactively explore metrics across time.
  • Exploratory analysis: Quickly test variable relationships with scatter plots, box plots, etc.
  • Teaching/Presenting: Embed responsive charts in web slides or notebooks.
  • Data storytelling: Combine multiple linked views to guide a narrative.

Backlinks