Specify the color

[robalvsan]

mv_dict = {'type': ['values', 'values'], 'quantity': [5, 10], 'category': ['mv', 'no_mv']}
df = pd.DataFrame(mv_dict)
alt.Chart(df).mark_bar().encode(
    y='type',
    x='sum(quantity)',
    color='category',
)

With this piece of code, the graph is generated automatically using blue for 'mv' and orange for 'no_mv'

Since in some contexts, the color has an implicit connotation...could it be possible to specify what color I want to use for 'mv' (e.g. red) and what for 'no_mv' (e.g. green)?

[afonit]

Here you go:

import altair as alt
import pandas as pd
mv_dict = {'type': ['values', 'values'], 'quantity': [5, 10], 'category': ['mv', 'no_mv']}
df = pd.DataFrame(mv_dict)
alt.Chart(df).mark_bar().encode(
    y='type',
    x='sum(quantity)',
    color=alt.Color('category',
                   scale=alt.Scale(
            domain=['mv', 'no_mv'],
            range=['red', 'green'])))

[robalvsan]

Awesome, thanks!

[jakevdp]

You can also use alt.Color('category', scale=alt.Scale(scheme=scheme_name)) where scheme_name is a string that matches any of the available Vega color schemes: https://vega.github.io/vega/docs/schemes/#reference

Edit: scheme should be a scale argument

[afonit]

@jakevdp , I tried what you posted:
alt.Color('category', scheme='accent')
I got a schema validation error:
Additional properties are not allowed ('scheme' was unexpected)

however doing:
color=alt.Color('category', scale=alt.Scale(scheme='accent'))
did work.
I updated to 2.1 to test what you had posted - perhaps I am doing something wrong.

[jakevdp]

sorry, yes, scheme is a Scale argument. glad you figured it out 😄

[sophiamaedler]

@jakevdp Do you have any tips on using vega schemes with log scales? I tried
alt.Color('Name', type = 'quantitative', scale=alt.Scale(scheme=scheme_name, type='log'))
but this resulted in the scheme command simply being ignored to generate the log scale and the standard colors being utilized.

[lightalchemist]

@Artemis-R I have the same issue as you but for the case where I set type="log". Do you have a fix? Thanks.

[SuperShinyEyes]

Here you go:

import altair as alt
import pandas as pd
mv_dict = {'type': ['values', 'values'], 'quantity': [5, 10], 'category': ['mv', 'no_mv']}
df = pd.DataFrame(mv_dict)
alt.Chart(df).mark_bar().encode(
    y='type',
    x='sum(quantity)',
    color=alt.Color('category',
                   scale=alt.Scale(
            domain=['mv', 'no_mv'],
            range=['red', 'green'])))

The specified colors break if you concatenate charts. :\

[fredgi123]

Is there a way to specifically have the 0 value returns a white color and the rest in the normal scheme?
If for instance, I use color=("variable:Q", scale=alt.Scale(scheme="blues")) my 0 values are still displaying blue values not white. thank you!

[tiaaburton]

@fredgi123 this answer on another issue was helpful. I also tried blues in an attempt to get 0 to display as white. I just used this discrete method to describe the scale I'd like to see. If you don't mind what the middle values are, you only have to specify 0, the cutoff for white coloring, and max. I added in the clamp=True param to make sure they weren't interacting with one another. #2258 (comment)