ERNIE/visualization.py at master · robeld/ERNIE · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.ticker import PercentFormatter
from matplotlib import colors

import seaborn
seaborn.set_context(context="paper")

def compareDistributions(arr1, arr2, path=None, show_fig=True, plot_title="linear initialization",
    arr1Title="Initial Centroids", arr2Title="Post K-means Centroids"):
    """ Graph for comparing distributions of two Numpy arrays. Used in this codebase to compare initial
    centroid values to those after kmeans is ran. Adapted from https://matplotlib.org/gallery/statistics/hist.html

    Args:
        arr1 (ndarray): 1d numpy array of floats, set to initial kmeans centroids in our code.
        arr2 (ndarray): 1d numpy array of floats, set to post kmeans centroids in our code.
            path (String): path to save graph
        show_fig (bool): Flag to show figure.
        plot_title (String): title of graph. Set to init method in our code.
        arr1Title (String): title for first subplot.
        arr2Title (String): title for second subplot.
    """
    figure, axs = plt.subplots(1, 2, tight_layout=True)#, sharey=True)
    figure.suptitle(plot_title, fontsize=12, y=0.03)

    w = np.ones_like(arr1)/float(len(arr1))
    N, bins, patches = axs[0].hist(arr1, density=False, weights=w)
    axs[0].set_title(arr1Title)
    fracs = N / N.max()
    norm = colors.Normalize(fracs.min(), fracs.max())
    for tf, tp in zip(fracs, patches):
        color = plt.cm.winter(norm(tf))
        tp.set_facecolor(color)
    axs[0].yaxis.set_major_formatter(PercentFormatter(xmax=1))

    figure.subplots_adjust(hspace=0.5)

    w_2 = np.ones_like(arr2)/float(len(arr2))
    N_2, bins_2, patches_2 = axs[1].hist(arr2, density=False, weights=w_2)
    axs[1].set_title(arr2Title)
    fracs_2 = N_2 / N_2.max()
    norm_2 = colors.Normalize(fracs_2.min(), fracs_2.max())
    for tf_2, tp_2 in zip(fracs_2, patches_2):
        color_2 = plt.cm.winter(norm(tf_2))
        tp_2.set_facecolor(color_2)
    axs[1].yaxis.set_major_formatter(PercentFormatter(xmax=1))

    if path:
        figure.savefig(path)
    if show_fig:
        plt.show()

def graphCDF(weights, path=None, title=None, show_fig=True, plot_title="Weights CDF Estimate", bins_factor=1):
    """
    Graph for graphing CDF of the values in a numpy array. Used in this codebase to graph
    CDF of weights of a layer. Adapted from https://matplotlib.org/examples/statistics/histogram_demo_cumulative.html

    Args:
        weights (ndarray): 1d numpy array of floats, set to weights of a layer in our code.
        path (String): path to save graph
        title (String): title of graph. Set to layer name in our code.
        show_fig (bool): Flag to show figure.
        plot_title (String): title of graph. Set to "Weights CDF Estimate" in our code.
        bins_factor (int): factor for the number of bins in the CDF histogram. More bins = less smooth curve.
    """
    figure, ax = plt.subplots(figsize=(8, 4))
    figure.suptitle(plot_title, fontsize=12)
    num_bins = int(weights.size/bins_factor)
    ax.hist(weights, bins=num_bins, density=1, histtype='step', cumulative=True, label='weights')
    if path:
        figure.savefig(path)
    if show_fig:
        plt.show()


if __name__=="__main__":
    x = np.random.randn(500)
    y = np.random.randn(7000)
    compareDistributions(x, y)
    graphCDF(x)