Extracting Vega Lite Transformations¶

In this notebook, we use extract out the transformations in the Vega Lite spec generated by Altair and push them to the server, by adding them to the SQL query with Ibis.

In [1]:

import altair as alt
import ibis.mapd
import IPython.display 
import ibis_vega_transform

Carrier names¶

First we connect to the table using Ibis:

In [2]:

conn = ibis.mapd.connect(
    host='metis.mapd.com', user='mapd', password='HyperInteractive',
    port=443, database='mapd', protocol= 'https'
)
t = conn.table("flights_donotmodify")

Then we compose an Altair chart using an ibis expression.

In [3]:

c = alt.Chart(t[t.carrier_name]).mark_bar().encode(
    x='carrier_name',
    y='count()'
)

Finally, we enable rendering that extracts the aggregate expressions and adds them onto the Ibis expresion:

In [4]:

Out[4]:

alt.Chart(...)

The only data loaded into the browser for this chart is one row for each carrier, because the counting transformation is pushed to the SQL statement.

Delay by Month¶

In [5]:

delay_by_month = alt.Chart(t[t.flight_dayofmonth, t.flight_month, t.depdelay]).mark_rect().encode(
    x='flight_dayofmonth:O',
    y='flight_month:O',
    color='average(depdelay)'
)
delay_by_month

Out[5]:

alt.Chart(...)

Interactive support: grouping by month¶

Now let's use widgets to facet this by month:

In [6]:

slider = alt.binding_range(min=1, max=12, step=1)
select_month = alt.selection_single(fields=['flight_month'],
                                   bind=slider, init={'flight_month': 1})

alt.Chart(t[t.flight_dayofmonth, t.depdelay, t.flight_month]).mark_line().encode(
    x='flight_dayofmonth:O',
    y='average(depdelay)'
).add_selection(
    select_month
).transform_filter(
    select_month
)

Out[6]:

alt.Chart(...)

Combing Tables: Comparing with Political Contributions¶

Now let's make a graph that also uses data from another table. Let's see how average delay time for flights from a state relates to the average political contribution size.

In [7]:

t_contributions = conn.table("contributions_donotmodify")

In [8]:

amount_by_state = alt.Chart(
    t_contributions[["contributor_state", "amount"]]
).mark_circle().encode(
    x="contributor_state",
    y=alt.Y("mean(amount):Q", axis=alt.Axis(grid=False)),
    color=alt.Color(value="green"),
    size="count()"
)

delay_by_state = alt.Chart(
    t[["origin_state", "depdelay"]]
).mark_square().encode(
    x="origin_state",
    y=alt.Y(
        "mean(depdelay):Q",
        axis=alt.Axis(grid=False)
    ),
    color=alt.Color(value="firebrick"),
    size="count()"
)

combined = (amount_by_state + delay_by_state).resolve_scale(y='independent', size='independent')
combined

Out[8]:

alt.LayerChart(...)

In [ ]: