Declarative Visualizations using Altair

My take on it: Fast explorative visualizations if you do not need to control the details

https://altair-viz.github.io/

Based on the idea that there is a grammar for graphics

https://www.amazon.com/Grammar-Graphics-Statistics-Computing/dp/0387245448

This notebook on Colab



In [1]:

    
import altair as alt

# remove when running on Colab or JupyterLab / add when running in a jupyter notebook
alt.renderers.enable('notebook')









    Out[1]:





RendererRegistry.enable('notebook')

Car insurance customer data



In [2]:

    
# Choose one of the two following data sets, the larger one gives better results, but might clutter the visualization depending on resolution

!curl -O https://raw.githubusercontent.com/DJCordhose/ai/master/notebooks/scipy/data/insurance-customers-1500.csv
!curl -O https://raw.githubusercontent.com/DJCordhose/ai/master/notebooks/scipy/data/insurance-customers-300.csv









    



  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed

  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0
100 26935  100 26935    0     0  78299      0 --:--:-- --:--:-- --:--:-- 78299
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed

  0     0    0     0    0     0      0      0 --:--:-- --:--:-- --:--:--     0
100  5421  100  5421    0     0  23166      0 --:--:-- --:--:-- --:--:-- 23166



In [3]:

    
import pandas as pd
# customers = pd.read_csv('./insurance-customers-300.csv', sep=';')
customers = pd.read_csv('./insurance-customers-1500.csv', sep=';')



In [4]:

    
customers.head()









    Out[4]:







  
    
      
      max speed
      age
      thousand km per year
      group
    
  
  
    
      0
      148.0
      44.0
      25.0
      1
    
    
      1
      168.0
      54.0
      24.0
      1
    
    
      2
      161.0
      26.0
      34.0
      0
    
    
      3
      137.0
      25.0
      10.0
      2
    
    
      4
      164.0
      38.0
      22.0
      1

A quick interactive overview



In [5]:

    
alt.Chart(customers).mark_point().encode(x='age', y='max speed', color='group')

A bit more elaborateed



In [6]:

    
# https://altair-viz.github.io/user_guide/encoding.html
# https://altair-viz.github.io/user_guide/customization.html

alt.Chart(customers).mark_point().encode(
    x='age',
    y='max speed',
    shape='group',
    color=alt.Color('group',
          type='nominal',
          legend=alt.Legend(title="Type of Driver by Color"),
          scale=alt.Scale(
                      range=['red', 'green', 'yellow'])),
).interactive()

Interactive filtering to explore the data



In [7]:

    
# https://altair-viz.github.io/gallery/interactive_layered_crossfilter.html

import altair as alt

brush = alt.selection(type='interval', encodings=['x'], resolve='intersect', empty='none')

# Define the base chart, with the common parts of the
# background and highlights
base = alt.Chart().mark_bar().encode(
    x=alt.X(alt.repeat('column'), type='quantitative'),
    y='count()'
).properties(
    width=250,
    height=150
)

# blue background with selection
background = base.properties(
    selection=brush
)

# yellow highlights on the transformed data
highlight = base.encode(
    color=alt.value('goldenrod')
).transform_filter(
    brush
)

# layer the two charts & repeat
alt.layer(
    background, highlight,
    data=customers
).repeat(
    column=["age", "max speed", "thousand km per year"]
)

We realize that a declarative API will necessarily be limited compared to the full programmatic APIs of Matplotlib, Bokeh, etc. That is a deliberate design choice we feel is needed to simplify the user experience of exploratory visualization.

https://github.com/altair-viz/altair#design-approach-and-solution



In [ ]:

	max speed	age	thousand km per year	group
0	148.0	44.0	25.0	1
1	168.0	54.0	24.0	1
2	161.0	26.0	34.0	0
3	137.0	25.0	10.0	2
4	164.0	38.0	22.0	1