BUG: groupby().plot(kind='scatter') broken

[jorisvandenbossche]

From SO: http://stackoverflow.com/questions/34123075/pandas-scatterplot-not-working-as-intended

import pandas as pd
url = 'https://raw.githubusercontent.com/cs109/2014_data/master/diamonds.csv'
diamonds = pd.read_csv(url, sep = ',', index_col=0)
diamonds.groupby('color').plot(kind='scatter', x='carat', y='price')

does not give you a scatter plot anymore (as an illustration, notebook with pandas 0.16.2 and pandas 0.17.0 )

This worked fine in pandas 0.16.2, but not anymore in 0.17.0 (or master). With using the new accessor syntax: diamonds.groupby('color').plot.scatter(x='carat', y='price') it still works.

Possibly related to the accessor plotting? @shoyer

cc @TomAugspurger @sinhrks