BUG: Discrepancy among SHAP beeswarm and Seaborn swarmplots

[ggmirandac]

Issue Description

I am using seaborn to plot the shap values and showing their relation with the data that was used for the prediction to avoid plotting nan values and I notice some differents between the plots. When I am compare both plots I notice that the the shap's api plot overstimates the value of the data.

Comparison:
SHAP plot:

Seaborn plot:

Just to inform, I am using the color map 'bwr' of matplotlib to compare.

Minimal Reproducible Example

import shap
# explain is the Explanation object
dic ={
    'feature_name':'ER',
    'shap_values':explain.values[:,0],
    'data': explain.data[:,0]
}
# for the seaborn
sns.swarmplot(x = 'shap_values', y = 'feature_name',hue = 'data', data = dic, palette='bwr', size = 3)

# for the shap
fig = plt.figure(figsize=(12, 5), dpi = 600)
shap.plots.beeswarm(explain[:,['ER']], show=False, max_display=len(explain.feature_names), plot_size=(6,10),
                    color_bar_label = '-',  color_bar=True, color = 'bwr')

Traceback

No response

Expected Behavior

There should be a correspondence between the values of the data in the plots. The color maps should show similar color behavior

Bug report checklist

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest release of shap.
• I have confirmed this bug exists on the master branch of shap.
• I'd be interested in making a PR to fix this bug

Installed Versions

version: 0.44.0