3
    plt.figure(figsize=(8, 8))

    sns.heatmap(conf_matrix, annot=True, fmt="d");
    plt.title("Confusion matrix")
    plt.ylabel('True class')
    plt.xlabel('Predicted class')

    html_fig = mpld3.fig_to_html(plt,template_type='general')
    plt.close(plt)

Code in HTML file to fetch the image :

    <div id="fig_container">
    {{ div_figure|safe }}
</div>

Using latest versions of Python and Django. On executing the attribute error is displayed

module 'matplotlib.pyplot' has no attribute 'canvas'

I am new to it and unable to resolve the error. Please help!!!

Sociopath
  • 13,068
  • 19
  • 47
  • 75
AmSri
  • 31
  • 1
  • 1
  • 2
  • 1
    Please add a few lines more context around that code snippet, particularly your import of pyplot and your use of canvas. – David Goldfarb Mar 18 '19 at 08:07
  • It seems you call `mpld3.fig_to_html(matplotlib.pyplot)` which is not correct. That function would probably expect a figure as input, `fig = plt.figure(); ....; mpld3.fig_to_html(fig)` – ImportanceOfBeingErnest Mar 18 '19 at 11:03

1 Answers1

3

I had a similar issue. You probably have solved the problem by now, but anyway I'm posting an answer.

The main point here is that mpld3 works with the fig object used in matplotlib. Like this:

fig, ax = plt.subplots()
ax.plot([1, 2, 3, 4])
html_fig = mpld3.fig_to_html(fig)

So for your solution you should get the figure object from your seaborn plot. It can be done either this way (whatch the ax argument):

fig, ax = plt.subplots()
sns.heatmap(conf_matrix, annot=True, fmt="d", ax=ax)
html_fig = mpld3.fig_to_html(fig)

Or this way:

fig = sns.heatmap(conf_matrix, annot=True, fmt="d").get_figure()
html_fig = mpld3.fig_to_html(fig)

If you want to include seaborn plots and matplotlib in general in django projects using mpld3 I'd recommend to view this question

Manuel Montoya
  • 1,206
  • 13
  • 25