Generate matrix plot for confusion matrix with pretty annotations.

plot_confusion_matrix.png

plot_confusion_matrix.py

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.metrics import confusion_matrix

def cm_analysis(y_true, y_pred, filename, labels, ymap=None, figsize=(10,10)):
    """
    Generate matrix plot of confusion matrix with pretty annotations.
    The plot image is saved to disk.
    args: 
      y_true:    true label of the data, with shape (nsamples,)
      y_pred:    prediction of the data, with shape (nsamples,)
      filename:  filename of figure file to save
      labels:    string array, name the order of class labels in the confusion matrix.
                 use `clf.classes_` if using scikit-learn models.
                 with shape (nclass,).
      ymap:      dict: any -> string, length == nclass.
                 if not None, map the labels & ys to more understandable strings.
                 Caution: original y_true, y_pred and labels must align.
      figsize:   the size of the figure plotted.
    """
    if ymap is not None:
        y_pred = [ymap[yi] for yi in y_pred]
        y_true = [ymap[yi] for yi in y_true]
        labels = [ymap[yi] for yi in labels]
    cm = confusion_matrix(y_true, y_pred, labels=labels)
    cm_sum = np.sum(cm, axis=1, keepdims=True)
    cm_perc = cm / cm_sum.astype(float) * 100
    annot = np.empty_like(cm).astype(str)
    nrows, ncols = cm.shape
    for i in range(nrows):
        for j in range(ncols):
            c = cm[i, j]
            p = cm_perc[i, j]
            if i == j:
                s = cm_sum[i]
                annot[i, j] = '%.1f%%\n%d/%d' % (p, c, s)
            elif c == 0:
                annot[i, j] = ''
            else:
                annot[i, j] = '%.1f%%\n%d' % (p, c)
    cm = pd.DataFrame(cm, index=labels, columns=labels)
    cm.index.name = 'Actual'
    cm.columns.name = 'Predicted'
    fig, ax = plt.subplots(figsize=figsize)
    sns.heatmap(cm, annot=annot, fmt='', ax=ax)
    plt.savefig(filename)

It's the first time I see this kind of oversized figure. Can you share the data you used (or some fake data) that can reproduce this?

PFA the data at https://drive.google.com/open?id=12Gx_O0Sjn0xocGQ-2jvW7Z03-61my29i

Thanks for prompt response.

Hi Hitvoice - this is awesome! Thanks so much posting this. I'm trying it out and see an odd thing with my binary classification CM. I've posted your version and what I get from using confusion_matrix. You can see that the classes are transposed and I'm wondering if I'm doing something wrong. I'd like to keep this aligned to what confusion_matrix displays so I thought I'd ask.

Hi Hitvoice - this is awesome! Thanks so much posting this. I'm trying it out and see an odd thing with my binary classification CM. I've posted your version and what I get from using confusion_matrix. You can see that the classes are transposed and I'm wondering if I'm doing something wrong. I'd like to keep this aligned to what confusion_matrix displays so I thought I'd ask.

What do you mean by "the classes are transposed"? Do you mean the order of 0 and 1?

Do you mean the order of 0 and 1?

Yes, I was looking to align this with sklearn's confusion matrix for consistency. Perhaps a parameter could be passed to indicate the order since various references have a different convention for the axes.

You can see from the code that the matrix is indeed computed from sklearn's "confusion_matrix" function. How did you get the first figure? If you prefer that kind of style, you can reorder the dataframe columns by cm = cm[cm.columns[::-1]] before creating the plot.

@hitvoice, it's really appreciated work. how to changet the clf.classes_ to other, for instance my classess are from 0 to 8, but i want to change those labels to string.

@rajeshkalakoti pass the value of ymap (dict[Any,str]) to cm_analysis. You can configure the class names in ymap.

for instance, labels are like "class-1, class-2,class-3, class-4,class-5, class-6,class-7, class-8,classs-9", how to configure it?