dcpython/spectral.py at main · Mafarka70/dcpython · 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
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
# -*- coding: utf-8 -*-
"""
Created on Fri Jan 31 12:41:54 2025

@author: gjgan
"""

import numpy as np
import matplotlib.pyplot as plt
from scipy.spatial.distance import pdist, squareform
from scipy.sparse.csgraph import laplacian
from sklearn.datasets import make_blobs, make_circles
from sklearn.cluster import SpectralClustering
from ucimlrepo import fetch_ucirepo
from dcutil import createCM
from kmeans import kmeans2

def epsilonneighbor(dm, epsilon=None):
    n = dm.shape[0]
    if epsilon is None:
        epsilon = np.mean(dm)/5
    S = np.zeros((n,n))
    S[dm < epsilon] = 1
    return S

def gaussian(dm, delta=None):
    if delta is None:
        delta = np.mean(dm)/5
    S = np.exp(-np.square(dm/delta)/2)
    return S

def spectral(S, h=3):
    L = laplacian(S, normed=True)
    ev = np.linalg.eigh(L)
    ind = np.argsort(ev[0])
    U = np.asanyarray(ev[1][:,ind[0:h]], float)
    U = U/np.sqrt(1e-8+np.sum(np.square(U), axis=1, keepdims=True))
    return U

# synthetic data 1
centers = [[3, 3], [-3, -3], [3, -3]]
X, y = make_blobs(n_samples=300, centers=centers, cluster_std=1, random_state=1)

dm = squareform(pdist(X))

S = epsilonneighbor(dm)
U = spectral(S, h=4)
bcm, bcc, bov, biters = kmeans2(U, k=3)
cm2 = createCM(y, bcm)
print([bov, biters])
print(cm2)

fig, ax = plt.subplots(2, 2, figsize=(8, 8))
ax[0,0].scatter(X[:,0], X[:,1], color="black")
ax[0,1].scatter(X[:,0], X[:,1], color="black")
ind = np.where(S > 0)
for i in range(ind[0].shape[0]):
    ij = [ind[0][i], ind[1][i]]
    ax[0,1].plot(X[ij,0], X[ij,1], '-', color="grey")
ax[1,0].scatter(U[:,0], U[:,1], color="black")
for i in range(3):
    center = bcc[i,:]
    ax[1,0].plot(center[0], center[1], "^", markerfacecolor="white",
        markeredgecolor="black", markersize=15)
markers = ["x", "o", "+"]
for i in range(3):
    members = bcm == i
    ax[1,1].plot(X[members, 0], X[members, 1], markers[i], color="black")
ax[0,0].set_title("D")
ax[0,1].set_title("S")
ax[1,0].set_title("U")
ax[1,1].set_title("CM")
fig.savefig("blob1spectral.png", bbox_inches='tight', dpi=300)


S = gaussian(dm)
U = spectral(S, h=4)
bcm, bcc, bov, biters = kmeans2(U, k=3)
cm2 = createCM(y, bcm)
print([bov, biters])
print(cm2)

clustering = SpectralClustering(n_clusters=3,
        assign_labels='discretize',
        random_state=0).fit(X)
cm = createCM(y, clustering.labels_)
print(cm)

# circles

X, y = make_circles(n_samples=300, noise=0.05, factor=0.4, random_state=0)

dm = squareform(pdist(X))

S = epsilonneighbor(dm)
U = spectral(S, h=4)
bcm, bcc, bov, biters = kmeans2(U, k=2)
cm2 = createCM(y, bcm)
print([bov, biters])
print(cm2)

fig, ax = plt.subplots(2, 2, figsize=(8, 8))
ax[0,0].scatter(X[:,0], X[:,1], color="black")
ax[0,1].scatter(X[:,0], X[:,1], color="black")
ind = np.where(S > 0)
for i in range(ind[0].shape[0]):
    ij = [ind[0][i], ind[1][i]]
    ax[0,1].plot(X[ij,0], X[ij,1], '-', color="grey")
ax[1,0].scatter(U[:,0], U[:,1], color="black")
for i in range(2):
    center = bcc[i,:]
    ax[1,0].plot(center[0], center[1], "^", markerfacecolor="white",
        markeredgecolor="black", markersize=15)
markers = ["x", "o", "+"]
for i in range(2):
    members = bcm == i
    ax[1,1].plot(X[members, 0], X[members, 1], markers[i], color="black")
ax[0,0].set_title("D")
ax[0,1].set_title("S")
ax[1,0].set_title("U")
ax[1,1].set_title("CM")
fig.savefig("circle1spectral.png", bbox_inches='tight', dpi=300)

clustering = SpectralClustering(n_clusters=2,
        assign_labels='discretize',
        gamma = 20,
        random_state=0).fit(X)
cm = createCM(y, clustering.labels_)
print(cm)


# iris data

iris = fetch_ucirepo(id=53)
X = iris.data.features
y = iris.data.targets

dm = squareform(pdist(X))


S = gaussian(dm)
U = spectral(S, h=3)
bcm, bcc, bov, biters = kmeans2(U, k=3)
cm2 = createCM(y, bcm)
print([bov, biters])
print(cm2)

clustering = SpectralClustering(n_clusters=3,
        assign_labels='discretize',
        random_state=0).fit(X)
cm = createCM(y, clustering.labels_)
print(cm)