Why doesn't seaborn barplot retain the order of a groupby index?

Question

I'm trying to create a seaborn barplot of a groupby object. Here is an example:

data = {'ID':[1, 2, 3, 1, 2, 3], 
        'value':[10, 20, 30, 100, 200, 300], 
        'other': ['a', 'b', 'c', 'd', 'e','f']
       }
data = pd.DataFrame(data)

I want to use groupby() to group the data by ID, calculate the mean value, and sort in descending order by value:

grouped = data.groupby('ID').mean()
grouped = grouped.sort_values(by='value', ascending=False)
grouped

Now I want to plot this using seaborn's barplot:

sns.barplot(x=grouped.index, y=grouped['value'])

This is the result:

The plot retains the original order and ignores the fact that I reassigned grouped to change the order.

Why does this happen? I've tried so many different approaches, and it always yields the same results.

Answer 1

You can always force the order:

sns.barplot(x=grouped.index, y=grouped['value'], order=grouped.index)