Need to save pandas correlation Highlighted table (cmap Matplotlib) as png image

Question

Used this code to genrate corelation table:

df1.drop(['BC DataPlus', 'AC Glossary'], axis=1).corr(method='pearson').style.format("{:.2}").background_gradient(cmap=plt.get_cmap('coolwarm'), axis=1)

This is the table generated:

I cant find any way to save this table as image. Thank you.

Perhaps if you were to break up the giant chain of calls into a couple of lines? — Mad Physicist, Jul 15 '18 at 10:19
Related, not 100% dupe: https://stackoverflow.com/q/35634238/2988730 — Mad Physicist, Jul 15 '18 at 10:21
@MadPhysicist no figure handle is returned and thats why I am unable to save as image — Muhammad Wasif, Jul 15 '18 at 10:35
One of the objects in that chain has a figure handle in it. Read the API docs carefully and you'll find it. — Mad Physicist, Jul 15 '18 at 15:59
@MadPhysicist I strongly doubt that. There is no matplotlib figure involved here at all (unless you know something that many others don't, in which case it would be good to provide an answer here). — ImportanceOfBeingErnest, Jul 15 '18 at 22:51
The main point here is that the table has a HTML representation, but no png representation (i.e you cannot do `IPython.display.display_png(table)`) hence you need to use any of the numerous workarounds listed in the link above or in links in that link. Depending on how often you need to do that, just doing a screenshot and saving that as png might be the fastest option. — ImportanceOfBeingErnest, Jul 15 '18 at 23:12
Guys I still haven't reached any solution, So can you please help me out. Thank you so much — Muhammad Wasif, Jul 16 '18 at 15:50

unutbu · Accepted Answer · 2018-07-17T01:08:31.547

The question you pose is difficult to answer if taken literally. The difficulty stems from the fact that df.style.render() generates HTML which is then sent to a browser to be rendered as an image. The result may not be exactly the same across all browsers either.

Python is not directly involved in the generation of the image. So there is no straight-forward Python-based solution.

Nevertheless, the issue of how to convert HTML to png was raised on the pandas developers' github page and the suggested answer was to use phantomjs. Other ways (that I haven't tested) might be to use webkit2png or GrabzIt.

We could avoid much of this difficulty, however, if we loosen the interpretation of the question. Instead of trying to produce the exact image generated by df.style (for a particular browser), we could generate a similar image very easily using seaborn:

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

df = pd.DataFrame(np.random.random((6, 4)), columns=list('ABCD'))
fig, ax = plt.subplots()
sns.heatmap(df.corr(method='pearson'), annot=True, fmt='.4f', 
            cmap=plt.get_cmap('coolwarm'), cbar=False, ax=ax)
ax.set_yticklabels(ax.get_yticklabels(), rotation="horizontal")
plt.savefig('result.png', bbox_inches='tight', pad_inches=0.0)

If you don't want to add the seaborn dependency, you could use matplotlib directly though it takes a few more lines of code:

import colorsys
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

df = pd.DataFrame(np.random.random((6, 4)), columns=list('ABCD'))
corr = df.corr(method='pearson')

fig, ax = plt.subplots()
data = corr.values
heatmap = ax.pcolor(data, cmap=plt.get_cmap('coolwarm'), 
                    vmin=np.nanmin(data), vmax=np.nanmax(data))
ax.set_xticks(np.arange(data.shape[1])+0.5, minor=False)
ax.set_yticks(np.arange(data.shape[0])+0.5, minor=False)
ax.invert_yaxis()
row_labels = corr.index
column_labels = corr.columns
ax.set_xticklabels(row_labels, minor=False)
ax.set_yticklabels(column_labels, minor=False)

def _annotate_heatmap(ax, mesh):
    """
    **Taken from seaborn/matrix.py**
    Add textual labels with the value in each cell.
    """
    mesh.update_scalarmappable()
    xpos, ypos = np.meshgrid(ax.get_xticks(), ax.get_yticks())
    for x, y, val, color in zip(xpos.flat, ypos.flat,
                                mesh.get_array(), mesh.get_facecolors()):
        if val is not np.ma.masked:
            _, l, _ = colorsys.rgb_to_hls(*color[:3])
            text_color = ".15" if l > .5 else "w"
            val = ("{:.3f}").format(val)
            text_kwargs = dict(color=text_color, ha="center", va="center")
            # text_kwargs.update(self.annot_kws)
            ax.text(x, y, val, **text_kwargs)

_annotate_heatmap(ax, heatmap)
plt.savefig('result.png', bbox_inches='tight', pad_inches=0.0)

Need to save pandas correlation Highlighted table (cmap Matplotlib) as png image

1 Answers1