useful_graphs

This library supports to draw graph that matplotlib does not support. Currently we have confusion matrix.

Install

pip install useful_graphs

Examples

Confusion Matrix

We recommend you to use scikit-learn's confusion_matrix to calculate the confusion matrix. This library helps you only to draw the graph of it.

Plot

import numpy as np
import useful_graphs

data = np.array([
                    [10, 3, 4],
                    [0, 9, 2],
                    [3, 2, 10],
                ])
classes = ['label-1', 'label-2', 'label-3']
cm = useful_graphs.ConfusionMatrix(data, class_list=classes)
cm.plot(to_normalize=True, to_show_number_label=True)

Save figure

To save a figure of confusion matrix, specify path with savepath argument. At this time, figure will not be shown.

import numpy as np
import useful_graphs

data = np.array([
                    [10, 3, 4],
                    [0, 9, 2],
                    [3, 2, 10],
                ])
classes = ['label-1', 'label-2', 'label-3']
cm = useful_graphs.ConfusionMatrix(data, class_list=classes)
cm.plot(to_normalize=True, to_show_number_label=True, save_path="{path_to_figure.pdf}")

Options for ConfusionMatrix.plot()

• to_normalize (bool, optional): If true, normalize confusion matrix for each row. (default False)
• title (str, optional): If specified, set a title to a figure. If none, title is not set. (default None)
• to_show_colorbar (bool, optional): If true, show colorbar to a figure. (default True)
• colorbar_label (str, optional): If specified, show label next to colorbar. (default None)
• to_show_label (bool, optional): If true, show label in a figure. (default True)
• to_show_number_label (bool, optional): If true, show number of data under the label. (default False)
• cmap (matplotlib colormap object, optional): Set colormap. See matplotlib for available colormaps. (default matplotlib.pyplot.cm.Reds)
• save_path (str, optional): If specified, save graph to specified path and not show the graph. (default None)
• xlabel (str, optional): If specified, change xlabel. (default 'Predicted label')
• ylabel (str, optional): If specified, change ylabel. (default 'True label')
• text_font_size (int, optional): If specified change font size. (default 10)
• figsize (tuple of shape=(w, h), optional): If specified, change size of a figure to w, h in inches. It changes figsize parameter of matplotlib.figure.Figure. (default None)