Week10/week10_mon.py at master · UCRPSY231/Week10 · 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
import sys, csv
import numpy as np
import scipy as sp
import pandas as pd
import matplotlib.pyplot as plt
from scipy.spatial.distance import pdist, squareform
from scipy.cluster.hierarchy import linkage, dendrogram

# input parameters
data_file = 'data1.csv'
dim1 = 0  # what SV dimensions to plot
dim2 = 1

# read data into a numpy matrix
data_matrix = np.array(list(csv.reader(open(data_file, "r"), delimiter=","))).astype("float")

# create some labels for the data
feature_labels = []
item_labels = []
sv_labels = []
for i in range(data_matrix.shape[0]):
    item_labels.append("item" + str(i + 1))
for i in range(data_matrix.shape[1]):
    sv_labels.append("sv" + str(i + 1))
    feature_labels.append("f" + str(i + 1))

# heatmap of the raw data
fig0, ax0 = plt.subplots()
ax0.imshow(data_matrix, interpolation='none')
ax0.set_xticks(np.arange(data_matrix.shape[1]) + 0.5, minor=False)
ax0.set_yticks(np.arange(data_matrix.shape[0]) + 0.5, minor=False)
ax0.set_xticklabels(feature_labels, minor=False)
ax0.set_yticklabels(item_labels, minor=False)

# compute the SVD
U, s, V = np.linalg.svd(data_matrix)
print(s)

# plot U and V (can be thought of as row and column-wise principal components)
fig1, axarr1 = plt.subplots(1, 2)
for ax, mat, ytlabels in zip(axarr1, [U, V], [item_labels, feature_labels]):
    ax.imshow(mat, interpolation='none')
    ax.set_yticks(np.arange(V.shape[0]) + 0.5, minor=False)
    ax.set_yticklabels(ytlabels, minor=False)
    ax.set_xticks([])
    # plot singular values on top of imshow (equal to square roots of the eigenvalues of covariance matrix)
    ax.plot(s)

# dimensionality reduction using above specified dimensions
# eigenvectors (of covariance matrix) are given by columns in U
fig2, ax2 = plt.subplots()
ax2.plot(U[:, dim1], U[:,dim2], "o")
ax2.set_xlabel('dim1')
ax2.set_ylabel('dim2')
plt.suptitle('Dimensionality Reduction with First 2 PCs', fontweight='bold', fontsize=14)

# plot the cluster diagram
data_dist = pdist(data_matrix) # computing the distance
data_link = linkage(data_dist) # computing the linkage
fig3, ax3 = plt.subplots()
dendrogram(data_link, ax=ax3)
ax3.set_xlabel('Items')
ax3.set_ylabel('Distance')
plt.suptitle('Hierarchical clustering', fontweight='bold', fontsize=14)

# show all figures
plt.show()