Line graph of cumulative sums over time aggregated by categorical value

[jimmywan]

I generated this using a bunch of custom pandas pivot/melt/cumsum logic:

I'd like to use altair transforms instead, but I don't quite understand how to do so.

I tried repurposing this but got lost. Is there another/better place to look for docs/examples of what I'm trying to accomplish? Happy to try and contribute some update examples if that would be helpful.

[jakevdp]

It should look pretty much exactly like the example you linked to, except with a groupby in the window transform and an additional color encoding. Here's a short example:

import pandas as pd
import altair as alt

data = pd.DataFrame({
  'qtr': ['2016Q1', '2016Q2', '2016Q3', '2016Q4', '2017Q1', '2017Q2', '2017Q3', '2017Q4',
          '2016Q1', '2016Q2', '2016Q3', '2016Q4', '2017Q1', '2017Q2', '2017Q3', '2017Q4'],
  'company': ['A', 'A', 'A', 'A', 'A', 'A', 'A', 'A',
              'B', 'B', 'B', 'B', 'B', 'B', 'B', 'B'],
  'value': [5, 4, 5, 6, 4, 7, 8, 3,
            7, 8, 8, 9, 6, 5, 8, 7],
})

alt.Chart(data).transform_window(
    cumulative='sum(value)',
    sort=[{"field": "qtr"}],
    groupby=['company'],
).mark_line().encode(
    x='qtr:N',
    y='cumulative:Q',
    color='company:N',
)

[jimmywan]

Apologies if I didn't make it clear in the ask, I'm trying to plot multiple lines, one for each categorical value.

[jakevdp]

Ah, sorry I missed that. Answer updated.

[jimmywan]

tyvm!