I have calculated 95% confidence interval for my data frame. When adding it to the grouped bar plot using Seaborn, I received an error that yerr must be a scalar or a 1D or (2, n) array-like.
Linguistic Measure ML Models score ci-b ci-t se
0 a RF 0.855 0.763 0.946 0.091
1 b RF 0.830 0.763 0.898 0.067
2 c RF 0.830 0.796 0.864 0.034
3 a SVM 0.841 0.823 0.860 0.018
4 b SVM 0.831 0.812 0.851 0.019
5 c SVM 0.835 0.820 0.850 0.015
6 a MLP 0.837 0.801 0.874 0.036
7 b MLP 0.810 0.721 0.900 0.089
8 c MLP 0.798 0.702 0.894 0.096
I used the code below trying to add error bars to the grouped barplot.
# Draw a nested barplot
g = sns.catplot(
data=sample, kind="bar", #update df
x="Linguistic Measure", y="score", hue="ML Models",
yerr=sample.pivot(index='Linguistic Measure',columns='ML Models',values='se').values,
ci=None, palette="BuPu", alpha=.6, height=5)
g.despine(left=True)
g.set_axis_labels("", "score")
g.legend.set_title("")
Then I got error message below:
Without adding the error bars, the graph looked fine