11

I have 4 different df.hist(columns=, by=) that I would like to insert inside of a GridSpec(2, 2).

Each one of them looks like this :

enter image description here

Here is the code :

stuff = [df1, df2, df4, df3]
col = ['blue', 'orange', 'grey', 'green']
fig = plt.figure(figsize=(10,10))
gs = gridspec.GridSpec(2, 2)

for i in range(0, len(stuff)):
    ax = plt.subplot(gs[i])
    stuff[i].hist(column='quanti_var', by=stuff[i].quali_var, alpha=.5, color=col[i], ax=ax)

I'm having the following UserWarning :

C:\Anaconda3\lib\site-packages\pandas\tools\plotting.py:3234: UserWarning: To output multiple subplots, the figure containing the passed axes is being cleared
  "is being cleared", UserWarning)

and not the output I am looking for :

enter image description here

I tried several things including the use of SubplotSpec without success. Any idea ?

Thank you guys for lending me your neurons !

gowithefloww
  • 2,211
  • 2
  • 20
  • 31
  • 1
    This would be nice to be able to do, but unfortunately not possible right now I don't think (at least using this interface). Looking at `plotting.py`, pandas always wants to make its own subplots when doing grouped histograms. Hence, you're only getting the very last DataFrame in your figure... – Ajean Sep 03 '15 at 15:43

1 Answers1

2

The solution is to place the matplotlib Axes objects returned by pd.DataFrame.hist() into a figure with the desired layout. Unfortunately, placing new Axes objects into an existing Figure is a bit involved.

GridSpec layout

Creating the layout you want is not too complicated using nested matplotlib.gridspec.GridSpecs (see here for an example). Something like this.

import matplotlib.gridspec as gs

num_outer_columns = 2
num_outer_rows = 2
num_inner_columns = 2
num_inner_rows = 3

outer_layout = gs.GridSpec(num_outer_rows, num_outer_columns)
inner_layout = []

for row_num in range(num_outer_rows):
    inner_layout.append([])
    for col_num in range(num_outer_columns):
        inner_layout[row_num].append(
            gs.GridSpecFromSubplotSpec(
                num_inner_rows, 
                num_inner_columns, 
                outer_layout[row_num, col_num]
            )
        )

You can create subplots within this grid using ax = plt.subplot(inner_layout[outer_row_num][outer_col_num][inner_row_num, inner_col_num]) and store the properly positioned axs for later.

Copying Axes into an existing Figure

Your df.hist() calls will produce something like this:

In [1]: dframe.hist(column=x, by=y)
Out[1]:
array([[<matplotlib.axes._subplots.AxesSubplot object at 0x1189be160>,
        <matplotlib.axes._subplots.AxesSubplot object at 0x118e3eb50>],
       [<matplotlib.axes._subplots.AxesSubplot object at 0x118e74d60>,
        <matplotlib.axes._subplots.AxesSubplot object at 0x118ea8340>],
       [<matplotlib.axes._subplots.AxesSubplot object at 0x118e76d62>,
        <matplotlib.axes._subplots.AxesSubplot object at 0x118ea9350>]],
      dtype=object)

Now you just have to replace the ax objects positioned earlier using your inner_layout with the AxesSubplot objects returned above. Unfortunately there isn't a convenient ax.from_axes(other_ax) method to do this, so you'll have to copy the Axes returned by df.hist() over manually by following this answer.

Emerson Harkin
  • 889
  • 5
  • 13