How to format the last line segment based on a datetime column? - altair

Question

I'm trying to make the final segment of a line plot dashed to indicate incomplete data. From what I can tell I should be able to do this using a condition on strokeDash. However I can't figure out how to get the condition predicate to work using a datetime field.
alt.Chart(rates)
.mark_line(point=True)
.encode(
x=alt.X("start_date:T", scale=alt.Scale(nice="week")),
y="install_rate",
strokeDash=alt.condition(
f"datum.start_date > toDate({start_dates[-2].isoformat()})",
alt.value([5, 5]), # dashed line: 5 pixels dash + 5 pixels space
alt.value([0]), # solid line
)
)
This gives me an error:
Error: Illegal callee type: MemberExpression

Answer 1

You can fix the error you are encountering by making sure that pandas reads in the dates as a temporal data type:
import pandas as pd
import altair as alt
rates = pd.DataFrame({
'start_date': pd.to_datetime(['2022-05-06', '2022-05-13', '2022-05-19', '2022-05-25']),
'install_rate': [0.05, 0.06, 0.08, 0.09],
})
alt.Chart(rates).mark_line(point=True).encode(
x=alt.X("start_date:T"),
y="install_rate",
color=alt.condition(
f"datum.start_date > toDate('2022-05-19')",
alt.value('blue'),
alt.value('red')
)
)
However, as you can see the lines is not amenable to modifications via a condition. I think this is because it is considered a single continuous mark whereas the points are split up and can be changed individually.
You could group the line by creating a new separate field and grouping by it, which creates two separate lines.
rates['above_threshold'] = rates['start_date'] > '2022-05-13'
alt.Chart(rates).mark_line(point=True).encode(
x=alt.X("start_date:T"),
y="install_rate",
color='above_threshold')
However, that causes issues with the gap as you can see above. I think for your case the easiest might be to layer two charts with filter transforms:
base = alt.Chart(rates).encode(
x=alt.X("start_date:T"),
y="install_rate",
)
base.mark_line(strokeDash=[5, 5]).transform_filter(
f"datum.start_date > toDate('2022-05-19')"
) + base.mark_line().transform_filter(
f"datum.start_date < toDate('2022-05-20')"
)