How to set alt.selection_point to allow one and only one option to be selected at a time (similar to a radio button filter)

[chrisHassan]

The below code replicates the issue I am having, using version altair==5.1.2. The issue I am having is that a user can highlight multiple categories by pressing shift while selecting them with their pointer. I didn't have this issue when using an older version of altair altair==4.2.0 (I recently upgraded to the latest).

# now we get base per a position
chart_df = pd.DataFrame(dict(
    category=list('abcde')
    , n=list(range(1,6))
))

# note, alt.selection_single didn't work for me either, and also get a deprecation warning
# to use alt.selection_point
category_selected = alt.selection_point( 
    fields=['category'],
    empty=True,
    value=[{'category': 'e'}],
    resolve='global',
    clear=True
)                                     

category_chart_filter = alt.Chart(chart_df).mark_bar().encode(
     x=alt.X('category', sort='-y')
    ,y='n'
    ,color='category'
    ,opacity=alt.condition(category_selected, alt.value(1), alt.value(0.2))
).add_params(
     category_selected
).properties(
    width=500, height=250
)


category_chart_filter

The following approach would accomplish the behaviour I wanted in altair==4.2.0

category_selection = alt.selection_single(
    fields=['category'],
    empty='none',
    init={'category': 'e'}
)  

[ChiaLingWeng]

Hi, I use colab to reproduce this issue,
https://colab.research.google.com/drive/17urtiUvV13chxNIFurzM-wzNzbZBIBRQ?usp=sharing
but the multi-selection function is fine

[binste]

You can set toggle=False within alt.selection_point:

import altair as alt
import pandas as pd
# now we get base per a position
chart_df = pd.DataFrame(dict(
    category=list('abcde')
    , n=list(range(1,6))
))

# note, alt.selection_single didn't work for me either, and also get a deprecation warning
# to use alt.selection_point
category_selected = alt.selection_point( 
    fields=['category'],
    value=[{'category': 'e'}],
    toggle=False
)                                     

category_chart_filter = alt.Chart(chart_df).mark_bar().encode(
     x=alt.X('category', sort='-y')
    ,y='n'
    ,color='category'
    ,opacity=alt.condition(category_selected, alt.value(1), alt.value(0.2))
).add_params(
     category_selected
).properties(
    width=500, height=250
)


category_chart_filter

From the docstring of alt.selection_point:

    toggle : string or boolean (optional)
        Controls whether data values should be toggled (inserted or
        removed from a point selection) or only ever inserted into
        point selections.
        One of:

        * True (default): the toggle behavior, which corresponds to
          "event.shiftKey". As a result, data values are toggled
          when the user interacts with the shift-key pressed.
        * False: disables toggling behaviour; the selection will
          only ever contain a single data value corresponding
          to the most recent interaction.
        * A Vega expression which is re-evaluated as the user interacts.
          If the expression evaluates to True, the data value is
          toggled into or out of the point selection. If the expression
          evaluates to False, the point selection is first cleared, and
          the data value is then inserted. For example, setting the
          value to the Vega expression True will toggle data values
          without the user pressing the shift-key.

Does this solve it for you?

[chrisHassan]

Thank you both @ChiaLingWeng for generating the colab notebook showing the issue.

Thank you @binste for highlighting the toggle argument to give me the behaviour I wanted. Super happy :-)