Transform filter from selection of temporal data is empty

I’m trying to use selection_interval to show a histogram of values selected from a scatterplot. This works fine with quantitative data but fails with temporal data that has not been transformed.

import altair as alt
import pandas as pd
import numpy as np

# Make dataframe from datetimes, random values, and sequential counts
dates = pd.date_range(start='1-1-2017-01', periods=200, freq='H')
values = np.random.randint(0, 20, size=200)
count = range(len(dates))

data = pd.DataFrame({'dt': dates, 'values': values, 'count': count})

brush = alt.selection_interval(encodings=['x'])

points = alt.Chart(data).mark_point().encode(
# Unmodified datetime, counts, and transformed datatime data for the x-axis
    x='dt:T',
#     x='count:Q',
#     x=alt.X('dt:T', timeUnit='yearmonthdatehours', axis=alt.Axis(title='Transformed dt')),
    y='values:Q',
).properties(
    selection=brush
)

bars = alt.Chart(data).mark_bar().encode(
    x='values:Q',
    y='count()'
).transform_filter(
    brush.ref()
)

points | bars

Initial plot with temporal dt on the x-axis

Empty bar chart/histogram when I select data in the scatterplot

Works fine with quantitative counts on the x-axis

Also works when dt is transformed

[jakevdp]

It’s up to you: free to just report issues here. I’m happy to echo them upstream where it’s appropriate.

[joelostblom]

@gschivley I am going through Altair issues to find those that have been resolved and can be closed. It looks to me like this issue has been solved in the latest version of Altair so I am closing it, but please feel free to reopen and add a comment if there is something you don’t think is resolved yet.