-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathMain.py
191 lines (165 loc) · 9.17 KB
/
Main.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
import pyswarms as ps
import Orange
import GEOparse
import Utils as ut
from Orange.data.pandas_compat import table_from_frame
from Bio import Geo
import UtilsDataset as udt
import UtilsFactory
import UtilsClassifier
import UtilsPSO
import numpy
import AbstractLearner, SVMLearner, KMeansLearner
import Orange.evaluation.scoring
import sys
from PyQt5.QtWidgets import *
import MainWindow
import KMeansLearner
from sklearn.metrics import pairwise_distances_argmin_min, pairwise_distances_argmin
from sklearn.model_selection import KFold,cross_val_score,cross_val_predict
import random
from sklearn.model_selection import GridSearchCV
'''
OPINIAO PESSOAL DA UTILIZACAO DE CLUSTERING NA SELECCAO DE FEATURES
https://www.quora.com/How-can-I-use-k-means-for-feature-extraction
'''
def main():
'''
Assim era outra forma de fazer as coisas, mas o dataset nao esta a vir muito bem
dataset = GEOparse.get_GEO(geo="GDS360", destdir="./")
ds = ut.pandas_to_orange(dataset.table)
print(ds.X.shape)
'''
'''
DEFINICAO INICIAL DO DATASET E DO CLASSIFICADOR UTILIZADO PARA TREINO E PREVISAO,
E DE OUTROS DADOS RELEVANTES
'''
data = Orange.data.Table("./datasetExplore")
factory = UtilsFactory.UtilsFactory()
dataset = factory.getUtil(name=ut.DATASET).getDataset(data)
#print(dataset.dataset.X.shape)
#print(dataset.getDataset().X)
#print(data.domain)
#print(data.X[0][9470:])
#REFORMULACAO DO CONTEUDO DA TABLE ORANGE, PRESENTE NO OBJETO UTILS DATASET
dataset= ut.transformMatrixDatasetInCorrectFormat(dataset)
#print(dataset.getDataset().X.shape)
#print(dataset.getDataset().Y)
classificador = factory.getUtil(ut.CLASSIFIER).getClassifier(gamma=0.01, vizinhos=5) #CRIACAO DO CLASSIFICADOR
examplesPredict = [0,1,22,23]
examplesTraining = range(2,22)
svmLeaner1 =SVMLearner.SVMLearner(gamma=0.5)
learner = svmLeaner1.getLearner()
'''
TREINO E PREVISAO DO DATASET ORIGINAL
'''
#APLICACAO DE CROSS VALIDATION
dataset = ut.applyMinMaxScaler(dataset)
kFold = KFold(n_splits=4, shuffle=True)
print("Score e Previsoes Iniciais - Cross Validation\n")
result = cross_val_score(learner, dataset.getDataset().X, dataset.getDataset().Y, cv=kFold, scoring='r2')
print("Media Score:\t",result.mean())
predictions = cross_val_predict(learner, dataset.getDataset().X, dataset.getDataset().Y, cv=3)
print("Prediction:\t",predictions)
print("\nSVM\n")
learner.fit(dataset.getDataset().X[examplesTraining], dataset.getDataset().Y[examplesTraining])
listSamplesPredict = ut.getSpecificSamples(dataset, examplesPredict)
predictions = learner.predict(listSamplesPredict)
realValuesPredict = ut.getSpecificOutputsFromDataset(dataset, examplesPredict)
print(Orange.evaluation.scoring.confusion_matrix(realValuesPredict, predictions))
print(ut.print_results(realValuesPredict,predictions))
#EXPERIMENTACAO DE CLASSIFICACAO E PREDICT
#listSamplesPredict = ut.getSpecificSamples(dataset, examplesPredict)
#predictions = learner.fit(dataset.getDataset().X[examplesTraining],dataset.getDataset().Y[examplesTraining]).predict(listSamplesPredict)
#realValuesPredict = ut.getSpecificOutputsFromDataset(dataset, examplesPredict)
#print(Orange.evaluation.scoring.confusion_matrix(realValuesPredict, predictions))
#print(ut.print_results(realValuesPredict,predictions))
'''
APLICACAO DO KMEANS ALGORITHM
'''
print("\nLoading K-Means\n")
#IDENTIFICACAO DO MELHOR VALOR DE K (CLUSTERS), TENDO EM CONTA UMA GAMA DE CLUSTERS E O DATASET EM ANALISE
transposeDataset = ut.applyTranspostMatrix(dataset)
#bestValueofK = ut.getBestValueOfK(dataset)
transposeDataset = ut.applyMinMaxScaler(transposeDataset)
#print(transposeDataset.getDataset().X)
kMeansObject = ut.applyClustering(60, transposeDataset)
#print(kMeansObject.getLearner().cluster_centers_.shape)
#print(kMeansObject.getLearner().labels_.shape)
#print(numpy.argwhere(kMeansObject.getLearner().labels_ == 3))
arrayBestFeatures = ut.getBestValuesForCluster(2,kMeansObject,transposeDataset)
#ARRAY WITH RELEVANT FEATURES --> ARRAY WITH 0'S OR 1'S --> 1'S RELEVANT FEATURES
myArray = ut.createBinaryNumpyArrayWithReducedFeatures(arrayBestFeatures, dataset)
#GET DATASET WITH FEATURE REDUCTION --> AFTER APPLY KMEANS ALGORITHM
reducedDataset = ut.createCloneOfReducedDataset(dataset, myArray)
# listSamplesPredict = ut.getSpecificSamples(reducedDataset, examplesPredict)
# predictions = learner.fit(reducedDataset.getDataset().X[examplesTraining],reducedDataset.getDataset().Y[examplesTraining]).predict(listSamplesPredict)
# realValuesPredict = ut.getSpecificOutputsFromDataset(reducedDataset, examplesPredict)
# print(Orange.evaluation.scoring.confusion_matrix(realValuesPredict, predictions))
# print(ut.print_results(realValuesPredict,predictions))
#APLICACAO DE CROSS VALIDATION
reducedDataset = ut.applyMinMaxScaler(reducedDataset)
print("Score e Previsoes KMeans - Cross Validation")
kFold = KFold(n_splits=4, shuffle=True)
result = cross_val_score(learner, reducedDataset.getDataset().X, reducedDataset.getDataset().Y, cv=kFold, scoring='r2')
print("Media Score:\t",result.mean())
predictions = cross_val_predict(learner, reducedDataset.getDataset().X, reducedDataset.getDataset().Y, cv=3)
print("Prediction:\t",predictions)
print("\nSVM")
learner.fit(reducedDataset.getDataset().X[examplesTraining], reducedDataset.getDataset().Y[examplesTraining])
listSamplesPredict = ut.getSpecificSamples(reducedDataset, examplesPredict)
predictions = learner.predict(listSamplesPredict)
realValuesPredict = ut.getSpecificOutputsFromDataset(reducedDataset, examplesPredict)
print(Orange.evaluation.scoring.confusion_matrix(realValuesPredict, predictions))
print(ut.print_results(realValuesPredict,predictions))
'''
APLICACAO DO BINARY PSO
'''
print("\nLoading BPSO\n")
# '''
# ABERTURA DA APLICACAO
# app = QApplication(sys.argv)
# window = MainWindow.MainWindow(dataset.getDataset())
# '''
#DEFINICAO DO ALGORITMO PSO
n_particles = 50
psoArgs = {UtilsPSO.UtilsPSO.INERCIA: 0.9, UtilsPSO.UtilsPSO.C1 : 1.4, UtilsPSO.UtilsPSO.C2 : 1.4, UtilsPSO.UtilsPSO.ALPHA : 0.80, UtilsPSO.UtilsPSO.NEIGHBORS : n_particles, 'p': 2} #p não é relevante, visto que todas as particulas se veem umas as outras, o p representa a distancia entre cada uma das particulas
psoAlgorithm = factory.getUtil(ut.PSO).getPso(**psoArgs)
optionsPySwarms = {'c1' : psoAlgorithm.getC1(), 'c2' : psoAlgorithm.getC2(), 'w' : psoAlgorithm.getInercia(), 'k' : psoArgs.get(UtilsPSO.UtilsPSO.NEIGHBORS), 'p' : psoArgs.get('p')}
#dimensionsOfProblem = dataset.getDataset().X.shape[1] #FEATURES DO DATASET
dimensionsOfProblem = reducedDataset.getDataset().X.shape[1]
initPos = ut.createArrayInitialPos(n_particles,dimensionsOfProblem,dimensionsOfProblem-4) #COLOCANDO UM NUMERO BAIXO NO INIT_POS, PERCEBEMOS QUE AQUI A ACCURACY JA VAI ALTERANDO, POIS COMO EXISTEM POUCAS AMOSTRAS, A ACCURACY REVELA-SE QUASE SEMPRE IGUAL, MESMO TREINANDO DUAS AMOSTRAS COM ATRIBUTOS SEMELHANTES E OUTPUTS DISTINTOS
optimizer = ps.discrete.BinaryPSO(n_particles=n_particles, dimensions=dimensionsOfProblem, options=optionsPySwarms)
bestCost, bestPos = optimizer.optimize(psoAlgorithm.aplicarFuncaoObjetivoTodasParticulas, 200, dataset= reducedDataset, classifier=classificador, alpha=psoAlgorithm.getAlpha())
#CONTAGEM DE QUANTAS FEATURES SAO RELEVANTES
bestPos = ut.listToNumpy(bestPos)
newFeatures = numpy.count_nonzero(bestPos)
#print(newFeatures)
#CRIACAO DA COPIA
deepCopy = ut.createCloneOfReducedDataset(reducedDataset,bestPos)
print(deepCopy.getDataset().X.shape)
'''
PREVISAO FINAL DAS 4 AMOSTRAS QUE NAO FORAM CONSIDERADAS NO TREINO DA APLICACAO DOS ALGORITMOS ANTERIORES
'''
#TREINO E PREVISAO, APENAS COM AS FEATURES SELECCIONADAS
deepCopy = ut.applyMinMaxScaler(deepCopy)
kFold = KFold(n_splits=4, shuffle=True)
result = cross_val_score(learner, deepCopy.getDataset().X, deepCopy.getDataset().Y, cv=kFold, scoring='r2')
print("Media Score:\t",result.mean())
predictions = cross_val_predict(learner, deepCopy.getDataset().X, deepCopy.getDataset().Y, cv=3)
print("Real:\t",deepCopy.getDataset().Y)
print("Prediction:\t",predictions)
print(Orange.evaluation.scoring.confusion_matrix(deepCopy.getDataset().Y, predictions))
print(ut.print_results(deepCopy.getDataset().Y,predictions))
print("\nFinal Results: SVM\n")
listSamplesPredict = ut.getSpecificSamples(deepCopy, examplesPredict)
predictionsAfterFeatureSelection = learner.fit(deepCopy.getDataset().X[examplesTraining],deepCopy.getDataset().Y[examplesTraining]).predict(listSamplesPredict)
realValuesPredict = ut.getSpecificOutputsFromDataset(deepCopy, examplesPredict)
print(Orange.evaluation.scoring.confusion_matrix(realValuesPredict, predictionsAfterFeatureSelection))
print(ut.print_results(realValuesPredict,predictionsAfterFeatureSelection))
# '''
# FECHO DA APLICACAO
# sys.exit(app.exec())
# '''
if __name__== "__main__":
main()