-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathcluster.py
141 lines (133 loc) · 5 KB
/
cluster.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
import random
import numpy as np
from difflib import SequenceMatcher
import pusher
import os
from dotenv import load_dotenv
load_dotenv()
class Clustering():
def __init__(self, num_of_means, schema):
self.num_of_means = num_of_means
self.data = []
self.schema = schema
self.cluster = {}
self.pusher_client = pusher.Pusher(
app_id=os.environ["PUSHER_APP_ID"],
key=os.environ["PUSHER_KEY"],
secret=os.environ["PUSHER_SECRET"],
cluster=os.environ["PUSHER_CLUSTER"],
ssl=True
)
def addData(self, data):
self.data = data
def updateData(self, row):
self.data.append(row)
def generateDistanceMatrix(self, f, type):
D = np.zeros((len(self.data), len(self.data)))
if type == "INTERVAL":
high = self.data[0][f]
low = self.data[0][f]
for i in range(1, len(self.data)):
high = max(high, self.data[i][f])
low = min(low, self.data[i][f])
for i in range(len(self.data)):
for j in range(len(self.data)):
D[i][j] = np.abs(self.data[i][f] -
self.data[j][f])/(high-low)
elif type == "RATIO":
for i in range(len(self.data)):
for j in range(len(self.data)):
D[i][j] = np.abs(np.log10(self.data[i][f]) -
np.log10(self.data[j][f]))
elif type == "NOMINAL":
for i in range(len(self.data)):
for j in range(len(self.data)):
D[i][j] = SequenceMatcher(
None, self.data[i][f], self.data[j][f]).ratio()
elif type == "BINARY SYMMETRIC":
for i in range(len(self.data)):
for j in range(len(self.data)):
if self.data[i][f] != self.data[j][f]:
D[i][j] = 1
elif type == "BINARY ASYMMETRIC":
for i in range(len(self.data)):
for j in range(len(self.data)):
if self.data[i][f] != self.data[j][f]:
D[i][j] = 1
elif type == "ORDINAL":
Mf = self.data[0][f]
for i in range(1, len(self.data)):
Mf = max(Mf, self.data[i][f])
for i in range(len(self.data)):
for j in range(len(self.data)):
zif = (self.data[i][f]-1)/(Mf-1)
zjf = (self.data[j][f]-1)/(Mf-1)
D[i][j] = np.abs(zif - zjf)
return D
def generateMixedMatrix(self, attr_idxs):
DFULL = []
D = np.zeros((len(self.data), len(self.data)))
for i in range(len(attr_idxs)):
DFULL.append(self.generateDistanceMatrix(
attr_idxs[i], self.schema[attr_idxs[i]][1]))
for i in range(len(self.data)):
for j in range(len(self.data)):
num = 0
den = 0
for f in range(len(attr_idxs)):
if not (self.schema[attr_idxs[f]][1] == "BINARY ASYMMETRIC" and self.data[i][attr_idxs[f]] == self.data[j][attr_idxs[f]] == 0):
num += DFULL[f][i][j]
den += 1
if den != 0:
D[i][j] = num/den
return D
def findClusters(self, num_of_iter, attr_idxs):
D = self.generateMixedMatrix(attr_idxs)
final_li = []
fin_li = []
while num_of_iter > 0:
initial = {}
st = set()
while (len(st) < self.num_of_means):
val = random.randint(0, len(self.data)-1)
st.add(val)
list_1 = list(st)
for k in range(self.num_of_means):
initial.setdefault(list_1[k], [])
for k in range(self.num_of_means):
initial[list_1[k]] = [list_1[k]]
for i in range(len(self.data)):
# ls = []
f = list(initial.keys())[0]
ls = D[i][f]
min_dist_clust = f
for k in initial.keys():
if D[i][k] < ls:
min_dist_clust = k
if i not in initial[min_dist_clust]:
initial[min_dist_clust].append(i)
# print(initial)
#### PUSHER ####
# self.pusher_client.trigger(
# 'clustering',
# 'iter_cluster',
# initial
# )
fin_li.append(initial)
num_of_iter = num_of_iter-1
for z, l in initial.items():
cost = 0.0
for e in l:
cost += D[z][e]
final_li.append(cost)
r = np.argmin(final_li)
op = fin_li[r]
# print(final_li, op)
self.cluster = op
#### PUSHER ####
# self.pusher_client.trigger(
# 'clustering',
# 'final_cluster',
# op
# )
return op