Categorical heatmap hack - how to add row-titles?

Aim

I am looking to make a categorical heatmap, made up of individual strips with different categorical columns in a dataframe, somewhat similar to this figure: taken from this paper.

What I need help with

I cannot figure out how to add row labels to the heatmap.

Making the heatmap

I use mark_rect and only give it an x and a color. Then I remove the legend, vconcat and remove the spacing between the individual plots. The legend would provide the row label I am looking for but I have to remove it, otherwise I cannot put the individual heatmap strips together tightly.

Making the legends

For the legends, I do the same as for the heatmap, except I don’t give an x in encode(), just a color.

Configuring

Finally, I configure_concat to make the heatmap strips snugg and configure_view to remove the outline of the empty plots that each of the legends comes with.

import pandas as pd
import altair as alt
#### Generating some Data ####
source = pd.DataFrame(
    {'fruits':['apple', 'banana', 'orange'], 
     'shape':['round', 'long', 'round'], 
     'weight':[0.3, 0.1, 0.25]})
#### Making the Heatmap ####
heatmap = alt.vconcat(
    alt.Chart(source).mark_rect().encode(
        x = alt.X('fruits', axis=None),
        color = alt.Color('shape', legend=None),
    ),
    alt.Chart(source).mark_rect().encode(
        x = 'fruits',
        color = alt.Color('weight', legend=None),
    )
)
#### Making the Legends ####
legends = alt.vconcat(
    alt.Chart(source).mark_rect(
        fillOpacity=0, 
        strokeOpacity=0
    ).encode(
        color = alt.Color('shape'),
    ), 
     alt.Chart(source).mark_rect(
        fillOpacity=0, 
        strokeOpacity=0
    ).encode(
        color = alt.Color('weight'),
    ), 
)

#### Configuring the concat ####
(
    heatmap | legends
).configure_concat(spacing=0 #makes strips of heatmap snugg
).configure_view(strokeOpacity=0) #removes plot border for the empty plots with the legends

This gives me the following heatmap:

Adding Row labels

The row labels here should be 'shape' and 'weight' but I cannot get that to work. My idea was to use the title as a row label with orient='left' and angle=90 but that doesn’t seem to work.

alt.Chart(source, title="shape").mark_rect().encode(
        x = alt.X('fruits', axis=None),
        color = alt.Color('shape', legend=None),
).configure_title(
    orient='left',
    angle=90
)

I don’t think I understand what the angle argument does in configure_title, because it doesn’t seem to rotate the title text, whether I give orient or not. From the docstring of configure_title:

angle : float Angle in degrees of title and subtitle text.

Thanks

Thank you for all the awesomeness in altair, the library is a true delight!

[jakevdp]

Unless you reference the calculation in an encoding, it will not be used in the chart:

alt.Chart(example_data).mark_rect().encode(
        x = alt.X('fruits'),
        y = alt.Y('y:N', title=None),
        color = alt.Color('shape'),
    ).transform_calculate(
    y='"shape"'
)

[jakevdp]

One way to do this kind of thing is to use a fold transform, and then use either a row encoding or a y-encoding; something like this:

import altair as alt
import pandas as pd

source = pd.DataFrame(
    {'fruits':['apple', 'banana', 'orange'], 
     'shape':['round', 'long', 'round'], 
     'weight':[0.3, 0.1, 0.25]})

alt.Chart(source).transform_fold(
    ["shape", "weight"]
).mark_rect().encode(
    x='fruits:N',
    y='key:N',
    color=alt.Color('value:N', legend=None)
)

This only works if you’re OK with all values being represented by the same color scale.