Bokeh Grouped Bar Plot Of multiple groupby

Hi,

I am having trouble figuring out how to Group Bar plot the pandas.core.series.Series below.

Not sure if this is a bug or I am doing something wrong.

If I convert the series to a dict, Bokeh makes a nice plot, but I cannot figure out how to plot the series where each origin

does not have a cyl of it’s own.

Thanks in advance…

Jim

from bokeh.charts import Bar, output_file, show

from bokeh.sampledata.autompg import autompg as df

from bokeh.plotting import output_notebook

output_notebook()

dataToPlot =df.groupby([‘cyl’,‘origin’]).mpg.mean()

print dataToPlot

type(dataToPlot)

     cyl origin
3 3 20.550000
4 1 28.013043
2 28.106557
3 31.595652
5 2 27.366667
6 1 19.645205
2 20.100000
3 23.883333
8 1 14.963107
Name: mpg, dtype: float64

Out[12]:

pandas.core.series.Series

The result of your aggregation in pandas has left you with a series with a multi-level index. All of the values are there, but they are nested and printed as such.

You can prove this to yourself by doing

dataToPlot.reset_index()

If I had to guess, you want to add the call to reset_index to your chain of commands. IOW:

dataToPlot =df.groupby(by=[‘cyl’,‘origin’]).mpg.mean().reset_index()

While we’re talking about pandas and not bokeh, an alternative to line above might be (I didn’t check):

dataToPlot =df.groupby(by=[‘cyl’,‘origin’], as_index=False).mpg.mean()

Hope that helps,

-Paul

Thank you, Paul!!!

It was a big help.

Matplotlib easily takes the series with a multi-level index, but Bokeh is so much prettier.

If this is the only way of making this type of plot, possibly it should be included in the examples.

Thanks again… Got to go, I have a lot of plots to make.