How to set mapping to colors in seaborn.heatmap?

Question

I'm trying to plot a correlation matrix with sns.heatmap(). Specifically, the code I run is

corr = train.corr()
sns.heatmap(corr, cmap = "RdBu_r")

This returns following image:

What I want is to 0 to correspond to white, positive correlations to be in shades of red and negative to be in shades of blue. I went through the documentation of seaborn, however, I wasn't able to figure it out.

Is there a way how to set this in seaborn?

Answer 1

A more robust solution might be to use the center argument in heatmap to set the center for plotting divergent data.

A = np.random.normal(1, 3, [5,5])
sns.heatmap(A, center=0, cmap=sns.diverging_palette(220, 20, as_cmap=True))

Results in the following figure

Answer 2

Try playing with vmin and vmax attribute, as mentioned here