mark_rect handling handling of missing data

I’m trying to create a heat map with mark_rect() as described here.

I’ve attached the data as a csv to this issue, as well as a snippet of (hopefully) reproducible code as well as a snippet of what I expect it to look like using Seaborn (note the difference in handling of “missing (NaN) data”, it’s grey in Seaborn and ?? in Altair).

csv needed to run example: testdf.csv.zip

import seaborn as sns
sns.set(rc={'figure.figsize':(11.7,8.27)})
df = pd.read_csv('testdf.csv')

# Seaborn plot
dff = df.pivot('state','year','annual_rate')
sns.heatmap(dff,vmin=0,vmax=300)

In Altair:

Any ideas?

[jakevdp]

And from the question, it sounds like you want to impute the y-axis, not the color, correct?

[firasm]

Indeed - impute the y-axis was the solution. Thanks!