/
kmeans_lib.py
341 lines (247 loc) · 12.9 KB
/
kmeans_lib.py
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
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
import matplotlib.pyplot as plt
import numpy as np
import ClassFile
import distances_lib as dist
__author__ = 'MagnieAr'
# k_means_alloc_table
# Inputs : data projected in PCA space, original dataset, number of dimensions, number of iterations, number of clusters
def k_means_reduced(all_projected_matrix, distance, d, n, k, weighted_distance):
# Initialization of clusters - according to K parameter
allocation_table = initial_allocation(k)
print(allocation_table)
results = ClassFile.Results()
all_projected_clusters = create_projected_cluster(all_projected_matrix, allocation_table)
# Iterations
for i in range(0, n):
for c in range(1, k + 1):
cluster_results = ClassFile.ClusterResults()
for s in range(1, d + 1):
if distance == "euclidean":
cur_dist = dist.euclidean_distance(all_projected_matrix, all_projected_clusters, s, c, weighted_distance)
if distance == "correlation-pca":
cur_dist = dist.correlation_distance(all_projected_matrix, all_projected_clusters, s, c, weighted_distance)
if distance == "dwt":
cur_dist = dist.dwt_distance(all_projected_matrix, all_projected_clusters, s, c, weighted_distance)
cluster_results.add_commodity(s)
cluster_results.add_distance(cur_dist)
# this object contains all the desired distances
results.add_cluster_result(cluster_results)
# Update assignments
allocation_table = update_allocation(results, allocation_table) # change prev alloc table based on curr results
print(allocation_table)
results = ClassFile.Results() # reset results
all_projected_clusters = create_projected_cluster(all_projected_matrix, allocation_table)
return allocation_table, all_projected_clusters;
def k_means_pc(all_eigen_vectors, all_projected_matrix, distance, d, n, k, weighted_distance):
# Initialization of clusters - according to K parameter
allocation_table = initial_allocation(k)
print(allocation_table)
results = ClassFile.Results()
all_pc_clusters = create_pc_prototype(all_eigen_vectors, all_projected_matrix, allocation_table)
# Iterations
for i in range(0, n):
for c in range(1, k + 1):
cluster_results = ClassFile.ClusterResults()
for s in range(1, d + 1):
if distance == "euclidean":
cur_dist = dist.euclidean_distance_pc(all_eigen_vectors, all_projected_matrix,
all_pc_clusters, s, c, weighted_distance)
if distance == "correlation-pca":
cur_dist = dist.correlation_distance_pc(all_eigen_vectors, all_projected_matrix,
all_pc_clusters, s, c, weighted_distance)
if distance == "dwt":
# TODO
cur_dist = dist.dwt_distance_pc(all_eigen_vectors, all_projected_matrix,
all_pc_clusters, s, c, weighted_distance)
cluster_results.add_commodity(s)
cluster_results.add_distance(cur_dist)
# this object contains all the desired distances
results.add_cluster_result(cluster_results)
# Update assignments
print "New allocation"
allocation_table = update_allocation(results, allocation_table) # change prev alloc table based on curr results
results = ClassFile.Results() # reset results
all_pc_clusters = create_pc_prototype(all_eigen_vectors, all_projected_matrix, allocation_table)
print "Code book"
print(allocation_table)
return allocation_table, all_pc_clusters;
def create_pc_prototype(all_eigen_vectors, all_projected_matrix, allocation_table):
all_prototypes = []
c = len(allocation_table[0]) # Columns
r = len(allocation_table) # Rows
for j in range(0, c): # clusters
contribution_1 = []
contribution_2 = []
contribution_3 = []
first_c = []
second_c = []
third_c = []
for p in range(0, r): # samples within a cluster
index = allocation_table[p, j] # get sample number
if index != 0:
first_c.append(all_eigen_vectors[index-1][0][0])
contribution_1.append(all_projected_matrix[index-1].components[0].contribution)
second_c.append(all_eigen_vectors[index-1][0][1])
contribution_2.append(all_projected_matrix[index-1].components[1].contribution)
third_c.append(all_eigen_vectors[index-1][0][2])
contribution_3.append(all_projected_matrix[index-1].components[2].contribution)
projected_cluster = ClassFile.ProjectedSeries("def")
component_1 = ClassFile.Component(np.mean(contribution_1))
component_1.add_series(np.mean(first_c, axis=0))
projected_cluster.add_components(component_1)
component_2 = ClassFile.Component(np.mean(contribution_2))
component_2.add_series(np.mean(second_c, axis=0))
projected_cluster.add_components(component_2)
component_3 = ClassFile.Component(np.mean(contribution_3))
component_3.add_series(np.mean(third_c, axis=0))
projected_cluster.add_components(component_3)
all_prototypes.append(projected_cluster)
return all_prototypes;
# TODO
def k_means_whole_data(original_samples, d, n, k):
# Initialization of clusters - according to K parameter
allocation_table = initial_allocation(k)
print(allocation_table)
results = ClassFile.Results()
# Iterations
for n in range(0, n):
for c in range(1, k + 1):
cluster_results = ClassFile.ClusterResults()
for s in range(1, d + 1):
pearson_coeff =0
dist =0
cluster_results.add_commodity(s)
cluster_results.add_distance(dist)
# this object contains all the desired distances
results.add_cluster_result(cluster_results)
# Update assignments
print "New allocation"
allocation_table = update_allocation(results, allocation_table) # change prev alloc table based on curr results
results = ClassFile.Results() # reset results
all_projected_clusters = create_projected_cluster(0, allocation_table)
print "Code book"
print(allocation_table)
return allocation_table, all_projected_clusters;
def initial_allocation(k):
# items = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 0, 0, 0])
items = np.array([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 0, 0])
np.random.shuffle(items)
allocation_table = np.array([items[0:4], items[4:8], items[8:12], items[12:16]])
return allocation_table;
def print_distances_sample_to_cluster(results, commodities_number):
for s in range(0, commodities_number-1):
d_to_cluster = [(results.cluster_results[c].distances[s], c)
for c in range(0, 4)]
d_to_cluster.sort()
print(s+1, d_to_cluster)
return;
def print_distances_cluster_to_sample(results, commodities_number):
for c in range(0, 4):
d_to_cluster = [(results.cluster_results[c].distances[s], s+1)
for s in range(0, commodities_number)]
d_to_cluster.sort()
print(c, d_to_cluster)
return 0;
def update_allocation(results, allocation_table):
c = len(allocation_table[0]) # Columns
r = len(allocation_table) # Rows
clusters = ClassFile.Results()
for j in range(0, c): # initialisation
cluster_results = ClassFile.ClusterResults()
clusters.add_cluster_result(cluster_results)
for j in range(0, c): # clusters
for p in range(0, r): # samples within a cluster
index = allocation_table[p, j] # get sample number
if index != 0:
# Get all distances corresponding to index, associated with cluster number (0 to k-1)
d_to_cluster = [(results.cluster_results[i].distances[index-1], i)
for i in range(0, c)]
d_to_cluster.sort() # Sort the distance in ascending order
t = 0
# This line ensures that if a cluster is full (number of sample inside = c), we fill the next one
# based on d_to_cluster (which has just been sorted)
while clusters.cluster_results[d_to_cluster[t][1]].number >= c:
t += 1
# Store the new results in a cluster with available space
clusters.cluster_results[d_to_cluster[t][1]].add_commodity(index)
new_alloc_table = np.zeros((r, c), dtype=np.int)
for j in range(0, c):
for p in range(0, clusters.cluster_results[j].number):
new_alloc_table[p][j] = int(clusters.cluster_results[j].commodities[p])
return new_alloc_table;
def create_projected_cluster(all_projected_matrix, allocation_table):
all_projected_clusters = []
# TODO : refactor the code
c = len(allocation_table[0]) # Columns
r = len(allocation_table) # Rows
for j in range(0, c): # clusters
contribution_1 = []
contribution_2 = []
contribution_3 = []
first_c = []
second_c = []
third_c = []
for p in range(0, r): # samples within a cluster
index = allocation_table[p, j] # get sample number
if index != 0:
first_c.append(all_projected_matrix[index-1].components[0].series[0])
contribution_1.append(all_projected_matrix[index-1].components[0].contribution)
second_c.append(all_projected_matrix[index-1].components[1].series[0])
contribution_2.append(all_projected_matrix[index-1].components[1].contribution)
third_c.append(all_projected_matrix[index-1].components[2].series[0])
contribution_3.append(all_projected_matrix[index-1].components[2].contribution)
projected_cluster = ClassFile.ProjectedSeries("def")
component_1 = ClassFile.Component(np.mean(contribution_1))
component_1.add_series(np.mean(first_c, axis=0))
projected_cluster.add_components(component_1)
component_2 = ClassFile.Component(np.mean(contribution_2))
component_2.add_series(np.mean(second_c, axis=0))
projected_cluster.add_components(component_2)
component_3 = ClassFile.Component(np.mean(contribution_3))
component_3.add_series(np.mean(third_c, axis=0))
projected_cluster.add_components(component_3)
all_projected_clusters.append(projected_cluster)
return all_projected_clusters;
def cost_of_clustering(all_projected_clusters, all_projected_matrix, allocation_table, weighted_distance):
cost = np.zeros(4)
for i in range(0, 4):
for j in range(0, 4):
if allocation_table[j][i] != 0:
cost[i] += dist.dwt_distance(all_projected_matrix, all_projected_clusters,
allocation_table[j][i]-1, i, weighted_distance)
return cost;
def display_final_clusters(all_projected_clusters, all_projected_matrix, allocation_table):
plt.figure(0)
colors = ["r", "m", "g", "y", "b"]
for i in range(0, 4):
plt.subplot(2, 2, i+1)
plt.plot(all_projected_clusters[i].components[0].series[0], label='centroid')
plt.title('Cluster {0}'.format(i+1))
for j in range(0, 4):
if allocation_table[j][i] != 0:
plt.plot(all_projected_matrix[allocation_table[j][i] - 1].components[0].series[0],
'{0}--'.format(colors[j]),
label='cmdty {0}'.format(allocation_table[j][i]))
plt.legend(loc='upper right')
plt.show()
return;
def display_final_clusters_pc(all_eigen_vectors, all_pc_clusters, allocation_table, components_number):
plt.figure(0)
colors = ["r", "m", "g", "y", "b"]
for i in range(0, 4):
plt.subplot(2, 2, i+1)
plt.plot(all_pc_clusters[i].components[0].series[0], 'b', label='centroid, PC1')
for k in range(1, components_number):
plt.plot(all_pc_clusters[i].components[k].series[0], '{0}'.format(colors[k]), label='centroid, PC{0}'.format(k+1))
plt.title('Cluster {0}'.format(i+1))
for j in range(0, 4):
if allocation_table[j][i] != 0:
plt.plot(all_eigen_vectors[allocation_table[j][i] - 1][0][0],
'{0}--'.format(colors[j]),
label='cmdty {0}'.format(allocation_table[j][i]))
for k in range(1, components_number):
plt.plot(all_eigen_vectors[allocation_table[j][i] - 1][0][k],
'{0}--'.format(colors[j]))
plt.legend(loc='upper right')
plt.show()
return;