Percentage of bin in bar chart

[LSchueler]

Hi and a big thank you for altair!

I have a table which I want to visualize as a bar chart. On the X axis, I want to bin the values into the given intervals (they are strings). On the Y axis, I want to give the percentage of positive values with respect to the individual bin. As an example for the example table below, for the 51 - 100 bin I want to show 50%, as there are 2 out of 4 positive values. The examples I found always calculate the relative values with respect to the total number of values.
I am unsure what transform methods to use and how to use them to get the wanted results.
It would be really great if someone could give me a hint!

ind  value                                        bins
0    Negative                                    0 - 50
1    Negative                                    0 - 50
2    Negative                                    0 - 50
3    Negative                                    0 - 50
4    Negative                                    0 - 50
5    Negative                                    0 - 50
6    Positive                                  51 - 100
7    Positive                                  51 - 100
8    Negative                                  51 - 100
9    Negative                                  51 - 100
10   Positive                                    0 - 50
11   Positive                                    0 - 50
12   Positive                                    0 - 50
13   Positive                                    0 - 50
14   Positive                                    0 - 50
15   Positive                                    0 - 50

[joelostblom]

There is transform_aggregate, but there is not proportion/percentage operation. You might be able to put something together using transform_calculate, but I am not sure how on the top of my head. You could also use pandas like this: df.groupby('bins')['value'].value_counts(normalize=True).xs('Positive', level='value').reset_index().

[jakevdp]

You can do this by roughly following the Percent of Total example, along with some pre-processing of your data in a calculate transform. For example:

import pandas as pd
import io
import altair as alt

data = """\
ind  value                                        bins
0    Negative                                    0 - 50
1    Negative                                    0 - 50
2    Negative                                    0 - 50
3    Negative                                    0 - 50
4    Negative                                    0 - 50
5    Negative                                    0 - 50
6    Positive                                  51 - 100
7    Positive                                  51 - 100
8    Negative                                  51 - 100
9    Negative                                  51 - 100
10   Positive                                    0 - 50
11   Positive                                    0 - 50
12   Positive                                    0 - 50
13   Positive                                    0 - 50
14   Positive                                    0 - 50
15   Positive                                    0 - 50
"""
source = pd.read_csv(io.StringIO(data), delimiter='\s{2,}', engine='python')

alt.Chart(source).transform_calculate(
    is_positive=(alt.datum.value == "Positive")
).transform_joinaggregate(
    total_pos='sum(is_positive)',
    total='count()',
    groupby=['bins']
).transform_calculate(
    percent=alt.datum.total_pos / alt.datum.total
).mark_bar().encode(
    y='bins:N',
    x=alt.X('percent:Q', axis=alt.Axis(format='%'))
)

[LSchueler]

Thank you so much for helping me out!