|
1
|
|
|
"""!
|
|
2
|
|
|
|
|
3
|
|
|
@brief Examples of usage utils.
|
|
4
|
|
|
|
|
5
|
|
|
@authors Andrei Novikov ([email protected])
|
|
6
|
|
|
@date 2014-2016
|
|
7
|
|
|
@copyright GNU Public License
|
|
8
|
|
|
|
|
9
|
|
|
@cond GNU_PUBLIC_LICENSE
|
|
10
|
|
|
PyClustering is free software: you can redistribute it and/or modify
|
|
11
|
|
|
it under the terms of the GNU General Public License as published by
|
|
12
|
|
|
the Free Software Foundation, either version 3 of the License, or
|
|
13
|
|
|
(at your option) any later version.
|
|
14
|
|
|
|
|
15
|
|
|
PyClustering is distributed in the hope that it will be useful,
|
|
16
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
17
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
18
|
|
|
GNU General Public License for more details.
|
|
19
|
|
|
|
|
20
|
|
|
You should have received a copy of the GNU General Public License
|
|
21
|
|
|
along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
22
|
|
|
@endcond
|
|
23
|
|
|
|
|
24
|
|
|
"""
|
|
25
|
|
|
|
|
26
|
|
|
import pyclustering.utils as utils;
|
|
27
|
|
|
|
|
28
|
|
|
from pyclustering.cluster.agglomerative import agglomerative;
|
|
29
|
|
|
from pyclustering.samples.definitions import SIMPLE_SAMPLES;
|
|
30
|
|
|
|
|
31
|
|
|
import matplotlib.pyplot as plt;
|
|
32
|
|
|
|
|
33
|
|
|
|
|
34
|
|
|
def cluster_distances(path_sample, amount_clusters):
|
|
35
|
|
|
distances = ['euclidian', 'manhattan', 'avr-inter', 'avr-intra', 'variance'];
|
|
36
|
|
|
|
|
37
|
|
|
sample = utils.read_sample(path_sample);
|
|
38
|
|
|
|
|
39
|
|
|
agglomerative_instance = agglomerative(sample, amount_clusters);
|
|
40
|
|
|
agglomerative_instance.process();
|
|
41
|
|
|
|
|
42
|
|
|
obtained_clusters = agglomerative_instance.get_clusters();
|
|
43
|
|
|
|
|
44
|
|
|
print("Measurements for:", path_sample);
|
|
45
|
|
|
|
|
46
|
|
|
for index_cluster in range(len(obtained_clusters)):
|
|
47
|
|
|
for index_neighbor in range(index_cluster + 1, len(obtained_clusters), 1):
|
|
48
|
|
|
cluster1 = obtained_clusters[index_cluster];
|
|
49
|
|
|
cluster2 = obtained_clusters[index_neighbor];
|
|
50
|
|
|
|
|
51
|
|
|
center_cluster1 = utils.centroid(sample, cluster1);
|
|
52
|
|
|
center_cluster2 = utils.centroid(sample, cluster2);
|
|
53
|
|
|
|
|
54
|
|
|
for index_distance_type in range(len(distances)):
|
|
55
|
|
|
distance = None;
|
|
56
|
|
|
distance_type = distances[index_distance_type];
|
|
57
|
|
|
|
|
58
|
|
|
if (distance_type == 'euclidian'):
|
|
59
|
|
|
distance = utils.euclidean_distance(center_cluster1, center_cluster2);
|
|
60
|
|
|
|
|
61
|
|
|
elif (distance_type == 'manhattan'):
|
|
62
|
|
|
distance = utils.manhattan_distance(center_cluster1, center_cluster2);
|
|
63
|
|
|
|
|
64
|
|
|
elif (distance_type == 'avr-inter'):
|
|
65
|
|
|
distance = utils.average_inter_cluster_distance(cluster1, cluster2, sample);
|
|
66
|
|
|
|
|
67
|
|
|
elif (distance_type == 'avr-intra'):
|
|
68
|
|
|
distance = utils.average_intra_cluster_distance(cluster1, cluster2, sample);
|
|
69
|
|
|
|
|
70
|
|
|
elif (distance_type == 'variance'):
|
|
71
|
|
|
distance = utils.variance_increase_distance(cluster1, cluster2, sample);
|
|
72
|
|
|
|
|
73
|
|
|
print("\tDistance", distance_type, "from", index_cluster, "to", index_neighbor, "is:", distance);
|
|
74
|
|
|
|
|
75
|
|
|
|
|
76
|
|
|
def display_two_dimensional_cluster_distances(path_sample, amount_clusters):
|
|
77
|
|
|
distances = ['euclidian', 'manhattan', 'avr-inter', 'avr-intra', 'variance'];
|
|
78
|
|
|
|
|
79
|
|
|
ajacency = [ [0] * amount_clusters for i in range(amount_clusters) ];
|
|
80
|
|
|
|
|
81
|
|
|
sample = utils.read_sample(path_sample);
|
|
82
|
|
|
|
|
83
|
|
|
agglomerative_instance = agglomerative(sample, amount_clusters);
|
|
84
|
|
|
agglomerative_instance.process();
|
|
85
|
|
|
|
|
86
|
|
|
obtained_clusters = agglomerative_instance.get_clusters();
|
|
87
|
|
|
stage = utils.draw_clusters(sample, obtained_clusters, display_result = False);
|
|
88
|
|
|
|
|
89
|
|
|
for index_cluster in range(len(ajacency)):
|
|
90
|
|
|
for index_neighbor_cluster in range(index_cluster + 1, len(ajacency)):
|
|
91
|
|
|
if ( (index_cluster == index_neighbor_cluster) or (ajacency[index_cluster][index_neighbor_cluster] is True) ):
|
|
92
|
|
|
continue;
|
|
93
|
|
|
|
|
94
|
|
|
ajacency[index_cluster][index_neighbor_cluster] = True;
|
|
95
|
|
|
ajacency[index_neighbor_cluster][index_cluster] = True;
|
|
96
|
|
|
|
|
97
|
|
|
cluster1 = obtained_clusters[index_cluster];
|
|
98
|
|
|
cluster2 = obtained_clusters[index_neighbor_cluster];
|
|
99
|
|
|
|
|
100
|
|
|
center_cluster1 = utils.centroid(sample, cluster1);
|
|
101
|
|
|
center_cluster2 = utils.centroid(sample, cluster2);
|
|
102
|
|
|
|
|
103
|
|
|
x_maximum, x_minimum, y_maximum, y_minimum = None, None, None, None;
|
|
104
|
|
|
x_index_maximum, y_index_maximum = 1, 1;
|
|
105
|
|
|
|
|
106
|
|
|
if (center_cluster2[0] > center_cluster1[0]):
|
|
107
|
|
|
x_maximum = center_cluster2[0];
|
|
108
|
|
|
x_minimum = center_cluster1[0];
|
|
109
|
|
|
x_index_maximum = 1;
|
|
110
|
|
|
else:
|
|
111
|
|
|
x_maximum = center_cluster1[0];
|
|
112
|
|
|
x_minimum = center_cluster2[0];
|
|
113
|
|
|
x_index_maximum = -1;
|
|
114
|
|
|
|
|
115
|
|
|
if (center_cluster2[1] > center_cluster1[1]):
|
|
116
|
|
|
y_maximum = center_cluster2[1];
|
|
117
|
|
|
y_minimum = center_cluster1[1];
|
|
118
|
|
|
y_index_maximum = 1;
|
|
119
|
|
|
else:
|
|
120
|
|
|
y_maximum = center_cluster1[1];
|
|
121
|
|
|
y_minimum = center_cluster2[1];
|
|
122
|
|
|
y_index_maximum = -1;
|
|
123
|
|
|
|
|
124
|
|
|
print("Cluster 1:", cluster1, ", center:", center_cluster1);
|
|
125
|
|
|
print("Cluster 2:", cluster2, ", center:", center_cluster2);
|
|
126
|
|
|
|
|
127
|
|
|
stage.annotate(s = '', xy = (center_cluster1[0], center_cluster1[1]), xytext = (center_cluster2[0], center_cluster2[1]), arrowprops = dict(arrowstyle = '<->'));
|
|
128
|
|
|
|
|
129
|
|
|
for index_distance_type in range(len(distances)):
|
|
130
|
|
|
distance = None;
|
|
131
|
|
|
distance_type = distances[index_distance_type];
|
|
132
|
|
|
|
|
133
|
|
|
if (distance_type == 'euclidian'):
|
|
134
|
|
|
distance = utils.euclidean_distance(center_cluster1, center_cluster2);
|
|
135
|
|
|
|
|
136
|
|
|
elif (distance_type == 'manhattan'):
|
|
137
|
|
|
distance = utils.manhattan_distance(center_cluster1, center_cluster2);
|
|
138
|
|
|
|
|
139
|
|
|
elif (distance_type == 'avr-inter'):
|
|
140
|
|
|
distance = utils.average_inter_cluster_distance(cluster1, cluster2, sample);
|
|
141
|
|
|
|
|
142
|
|
|
elif (distance_type == 'avr-intra'):
|
|
143
|
|
|
distance = utils.average_intra_cluster_distance(cluster1, cluster2, sample);
|
|
144
|
|
|
|
|
145
|
|
|
elif (distance_type == 'variance'):
|
|
146
|
|
|
distance = utils.variance_increase_distance(cluster1, cluster2, sample);
|
|
147
|
|
|
|
|
148
|
|
|
print("\tCluster distance -", distance_type, ":", distance);
|
|
149
|
|
|
|
|
150
|
|
|
x_multiplier = index_distance_type + 3;
|
|
151
|
|
|
if (x_index_maximum < 0):
|
|
152
|
|
|
x_multiplier = len(distances) - index_distance_type + 3;
|
|
153
|
|
|
|
|
154
|
|
|
y_multiplier = index_distance_type + 3;
|
|
155
|
|
|
if (y_index_maximum < 0):
|
|
156
|
|
|
y_multiplier = len(distances) - index_distance_type + 3;
|
|
157
|
|
|
|
|
158
|
|
|
x_text = x_multiplier * (x_maximum - x_minimum) / (len(distances) + 6) + x_minimum;
|
|
159
|
|
|
y_text = y_multiplier * (y_maximum - y_minimum) / (len(distances) + 6) + y_minimum;
|
|
160
|
|
|
|
|
161
|
|
|
#print(x_text, y_text, "\n");
|
|
162
|
|
|
stage.text(x_text, y_text, distance_type + " {:.3f}".format(distance), fontsize = 9, color='blue');
|
|
163
|
|
|
|
|
164
|
|
|
plt.show();
|
|
165
|
|
|
|
|
166
|
|
|
|
|
167
|
|
|
def display_cluster_distances_simple_sample_01():
|
|
168
|
|
|
display_two_dimensional_cluster_distances(SIMPLE_SAMPLES.SAMPLE_SIMPLE1, 2);
|
|
169
|
|
|
|
|
170
|
|
|
def display_cluster_distances_simple_sample_02():
|
|
171
|
|
|
display_two_dimensional_cluster_distances(SIMPLE_SAMPLES.SAMPLE_SIMPLE2, 3);
|
|
172
|
|
|
|
|
173
|
|
|
def display_cluster_distances_simple_sample_03():
|
|
174
|
|
|
display_two_dimensional_cluster_distances(SIMPLE_SAMPLES.SAMPLE_SIMPLE3, 4);
|
|
175
|
|
|
|
|
176
|
|
|
|
|
177
|
|
|
def print_cluster_distances_simple_sample_07():
|
|
178
|
|
|
cluster_distances(SIMPLE_SAMPLES.SAMPLE_SIMPLE7, 2);
|
|
179
|
|
|
|
|
180
|
|
|
def print_cluster_distances_simple_sample_08():
|
|
181
|
|
|
cluster_distances(SIMPLE_SAMPLES.SAMPLE_SIMPLE8, 4);
|
|
182
|
|
|
|
|
183
|
|
|
|
|
184
|
|
|
display_cluster_distances_simple_sample_01();
|
|
185
|
|
|
display_cluster_distances_simple_sample_02();
|
|
186
|
|
|
display_cluster_distances_simple_sample_03();
|
|
187
|
|
|
|
|
188
|
|
|
print_cluster_distances_simple_sample_07();
|
|
189
|
|
|
print_cluster_distances_simple_sample_08();
|
|
190
|
|
|
|
|
191
|
|
|
|
annoviko /
pyclustering
