/
clasificacion.py
160 lines (117 loc) · 4.24 KB
/
clasificacion.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
# -*- coding: utf-8 -*-
"""
Created on Mon May 21 11:54:58 2018
@author: Javier Fumanal Idocin
"""
import numpy as np
import orangecontrib.associate.fpgrowth as fp
from clustering import clustering, filter_numerical
from sklearn.ensemble import IsolationForest
from eventregistry import EventRegistry, QueryArticlesIter
def extremos_incertidumbre(fit, datos, cluster):
'''
Devuelve la mayor y menor distancia de un conjunto de datos a un fit, dada una funcion
de distancia.
'''
maximo = 0
minimo = np.Inf
for dato in range(datos.shape[0]):
dist = distancia(fit, datos.iloc[dato].values.reshape(1,-1), cluster)
if dist > maximo:
maximo = dist
elif dist < minimo:
minimo = dist
return [maximo, minimo]
def calcular_incertidumbre(fit, dato, confianzas, cluster):
'''
Calculate the confidence in a prediction.
'''
minimo = confianzas[1]
maximo = confianzas[0]
dist = distancia(fit, dato, cluster)
confianza = (dist - minimo) / maximo
if(confianza < 0):
return 0
else:
return confianza
def distancia(fit, dato, cluster):
'''
Calcultes distances from points to clusters.
'''
return fit.transform(dato)[0][cluster]
def classify(datos, fit):
'''
Classifies data.
'''
return fit.predict(datos)
def tasa_aceptabilidad(incertidumbres):
'''
Gives the percentage of acceptable classifications made.
'''
incertidumbres2 = np.array(incertidumbres.copy())
incertidumbres2[np.array(incertidumbres2) < 1.0] = 1.0
incertidumbres2[np.array(incertidumbres2) > 1.0] = 0.0
return np.mean(incertidumbres2)
def transacciones_profundidad(X, profundidad = 1):
'''
'''
clusters = np.array(X['cluster'])
rows = len(clusters) - profundidad
cols = len(np.unique(X.cluster))
holder = np.zeros([rows, cols])
holder.fill('0')
for i in range(profundidad, rows):
holder[i-profundidad,clusters[i]] = 1
for j in range(i-profundidad,i):
holder[i - profundidad,clusters[j]] = 1
return holder
def rules_extractor(X, profundidades=range(4), metric = 0.3):
res = {}
for i in profundidades:
T = transacciones_profundidad(X,i)
itemsets = dict(fp.frequent_itemsets(T, metric))
rules = [(P, Q, supp, conf) for P, Q, supp, conf in fp.association_rules(itemsets, metric)]
res[i] = (itemsets, rules)
return res
def anomaly_detection(X, name = 'anomaly'):
pr = IsolationForest()
pr.fit(filter_numerical(X))
x = pr.predict(filter_numerical(X))
X[name] = x
X[name] = X[name].astype(str)
def noticias(theme, dates):
er = EventRegistry("c4b6b663-d180-4f4c-8163-4c45fbc7cbd7")
q = QueryArticlesIter(conceptUri = theme, dateStart = dates[0], dateEnd=dates[1] )
for art in q.execQuery(er, sortBy = "date"):
print(art)
def conjunto_segmentados(segmentados, n=3):
import pandas as pd
segments_df=pd.DataFrame()
lens = []
inicio=0
for conj in segmentados:
segments_df = segments_df.append(conj)
lens.append([inicio,conj.shape[0]])
inicio = conj.shape[0]
fit = clustering(segments_df, n)
segments_df['cluster'] = fit.labels_
segments_df['cluster'] = segments_df['cluster'].astype(str)
for x in range(len(lens)):
segmentados[x] = segments_df.iloc[lens[x]]
confidences = []
for i in range(n):
confidences.append(extremos_incertidumbre(fit, filter_numerical(segments_df), i))
return segmentados, fit, confidences, segments_df
def clustering_df(X, fit, confidences=None):
clusters = []
confs = []
X.drop('cluster', axis=1,inplace=True, errors='ignore')
for i in range(X.shape[0]):
pred = classify(X.iloc[i].values.reshape(1,-1), fit)
dis = distancia(fit, X.iloc[i].values.reshape(1,-1), pred)[0]
if not(confidences is None):
conf = (dis - confidences[pred[0]][1]) / confidences[pred[0]][0]
confs.append(conf)
clusters.append(pred[0])
X['cluster'] = clusters
return confs