-
Notifications
You must be signed in to change notification settings - Fork 0
/
brafClassification.py
331 lines (284 loc) · 12.9 KB
/
brafClassification.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
import pandas as pd
import numpy as np
from scipy import stats
from sklearn.model_selection import train_test_split
from sklearn import tree
from sklearn.metrics import confusion_matrix
from sklearn.neighbors import LocalOutlierFactor, KNeighborsClassifier
from sklearn.svm import SVC
from sklearn.ensemble import BaggingClassifier,AdaBoostClassifier
from joblib import dump, load
import os
import graphviz
import time
"""
Cuva model
"""
def save_model(clf, metod):
dump(clf, os.path.join('models', metod + '.joblib'))
"""
Nalazi nula kolone i eliminise ih u zavisnosti od odabranog nacina - unija ili presek
"""
def zero_columns_to_remove(df1, df2, removal_type):
first_zero_columns = df1.columns[(df1 == 0).all()]
second_zero_columns = df2.columns[(df2 == 0).all()]
if(removal_type == 'union'):
removed_cols = np.union1d(first_zero_columns, second_zero_columns)
elif (removal_type == 'intersection'):
removed_cols = np.intersect1d(first_zero_columns, second_zero_columns)
else:
print("Bad type of removing zero values")
return removed_cols
"""
Cuva trenutni CSV fajl
"""
def save_csv(df, filename):
df.to_csv(filename + '.csv', sep=',')
"""
Razdvaja klasu od podataka i vraca oba
"""
def seperate_data_class(df):
v_data = df.values[:, :-1]
v_class = df.values[:, -1:]
return v_data, v_class
"""
STABLO ODLUCIVANJA - klasifikacija
"""
def decisionTree(x_train, x_test, y_train, y_test, criteria, depth=None, generateGraph=[]):
print("---------------------------------------------\n")
print("Rezultati za stablo: [" + str(criteria) + " " + str(depth) + "] \n")
beginTime = time.time()
print("Kreiranje klasifikatora ... \n")
clf = tree.DecisionTreeClassifier(criterion=criteria, max_depth=depth)
clf.fit(x_train,y_train.ravel())
save_model(clf, "tree_" + str(criteria) + "_" + str(depth))
print('Trening tacnost: {}'.format(clf.score(x_train, y_train)))
print('Test tacnost: {}'.format(clf.score(x_test, y_test)))
y_predict_train = clf.predict(x_train)
y_predict_test = clf.predict(x_test)
print("Matrica kofuzije trening vrednosti: \n" + str(confusion_matrix(y_train, y_predict_train)))
print("Matrica kofuzije test vrednosti: \n" + str(confusion_matrix(y_test, y_predict_test)))
endTime = time.time()
elapsedTime = endTime - beginTime
print(f"Vreme potrebno za izvrsavanje: {elapsedTime:.4f} \n")
if generateGraph:
print("Kreiranje grafa")
dot_data = tree.export_graphviz(clf,
out_file=None,
feature_names= generateGraph,
class_names=['parental (BRAF inhibitor sensitive)', 'BRAF inhibitor resistant'])
graph = graphviz.Source(dot_data)
if(depth == None):
depth = "full"
graph.render("dtGraph/DecisionTree_" + criteria + "_" + str(depth))
"""
KNN - klasifikacija
"""
def knn(x_train, x_test, y_train, y_test, n_neigh, weights='uniform', algorithm = 'auto'):
print("---------------------------------------------\n")
print("Rezultati za knn: [" + str(n_neigh) + " " + str(weights) + " " + str(algorithm) + "] \n")
beginTime = time.time()
print("Kreiranje klasifikatora ... \n")
clf = KNeighborsClassifier(n_neigh, weights, algorithm)
clf.fit(x_train,y_train.ravel())
save_model(clf, "knn_" + str(n_neigh) + "_" + str(weights) + "_" + str(algorithm))
print('Trening tacnost: {}'.format(clf.score(x_train, y_train)))
print('Test tacnost: {}'.format(clf.score(x_test, y_test)))
y_predict_train = clf.predict(x_train)
y_predict_test = clf.predict(x_test)
print("Matrica kofuzije trening vrednosti: \n" + str(confusion_matrix(y_train, y_predict_train)))
print("Matrica kofuzije test vrednosti: \n" + str(confusion_matrix(y_test, y_predict_test)))
endTime = time.time()
elapsedTime = endTime - beginTime
print(f"Vreme potrebno za izvrsavanje: {elapsedTime:.4f} \n")
"""
SVM - klasifikacija
"""
def svm(x_train, x_test, y_train, y_test, kernel, gamma='auto'):
print("---------------------------------------------\n")
print("Rezultati za svm: [" + str(kernel) + " " + str(gamma) + "] \n")
beginTime = time.time()
print("Kreiranje klasifikatora ... \n")
clf = SVC(kernel=kernel, gamma=gamma)
clf.fit(x_train,y_train.ravel())
save_model(clf, "svm_" + str(kernel) + "_" + str(gamma))
print('Trening tacnost: {}'.format(clf.score(x_train, y_train)))
print('Test tacnost: {}'.format(clf.score(x_test, y_test)))
y_predict_train = clf.predict(x_train)
y_predict_test = clf.predict(x_test)
print("Matrica kofuzije trening vrednosti: \n" + str(confusion_matrix(y_train, y_predict_train)))
print("Matrica kofuzije test vrednosti: \n" + str(confusion_matrix(y_test, y_predict_test)))
endTime = time.time()
elapsedTime = endTime - beginTime
print(f"Vreme potrebno za izvrsavanje: {elapsedTime:.4f} \n")
"""
BAGGING - klasifikacija
"""
def bagging(x_train, x_test, y_train, y_test, n, classifier=None):
print("---------------------------------------------\n")
print("Rezultati za bagging: [" + str(classifier) + " " + str(n) + "] \n")
beginTime = time.time()
print("Kreiranje klasifikatora ... \n")
if(classifier == "svc"):
unit = SVC(kernel='poly', gamma='auto')
elif(classifier == "tree"):
unit = tree.DecisionTreeClassifier()
elif(classifier == "knn"):
unit = KNeighborsClassifier(3, algorithm="brute")
else:
unit = None
clf = BaggingClassifier(unit, n_estimators=n)
clf.fit(x_train,y_train.ravel())
save_model(clf, "bagging_" + str(classifier) + "_" + str(n))
print('Trening tacnost: {}'.format(clf.score(x_train, y_train)))
print('Test tacnost: {}'.format(clf.score(x_test, y_test)))
y_predict_train = clf.predict(x_train)
y_predict_test = clf.predict(x_test)
print("Matrica kofuzije trening vrednosti: \n" + str(confusion_matrix(y_train, y_predict_train)))
print("Matrica kofuzije test vrednosti: \n" + str(confusion_matrix(y_test, y_predict_test)))
endTime = time.time()
elapsedTime = endTime - beginTime
print(f"Vreme potrebno za izvrsavanje: {elapsedTime:.4f} \n")
"""
BOOSTING - klasifikacija
"""
def boosting(x_train, x_test, y_train, y_test, learning=1.0, classifier=None):
print("---------------------------------------------\n")
print("Rezultati za boosting: [" + str(classifier) + " " + str(learning) + "] \n")
beginTime = time.time()
print("Kreiranje klasifikatora ... \n")
if(classifier == "svc"):
unit = SVC(kernel='poly', gamma='auto')
alg = "SAMME"
elif(classifier == "tree"):
unit = tree.DecisionTreeClassifier()
alg = "SAMME.R"
else:
alg = "SAMME.R"
unit = None
clf = AdaBoostClassifier(unit, learning_rate=learning, algorithm = alg)
clf.fit(x_train,y_train.ravel())
save_model(clf, "boosting_" + str(classifier) + "_" + str(learning))
print('Trening tacnost: {}'.format(clf.score(x_train, y_train)))
print('Test tacnost: {}'.format(clf.score(x_test, y_test)))
y_predict_train = clf.predict(x_train)
y_predict_test = clf.predict(x_test)
print("Matrica kofuzije trening vrednosti: \n" + str(confusion_matrix(y_train, y_predict_train)))
print("Matrica kofuzije test vrednosti: \n" + str(confusion_matrix(y_test, y_predict_test)))
endTime = time.time()
elapsedTime = endTime - beginTime
print(f"Vreme potrebno za izvrsavanje: {elapsedTime:.4f} \n")
"""
Nalazi kolone van granica
"""
def find_outlier_columns(df):
outlier_res = (df.drop(columns="Class")).T
outliers = outlier_res[(np.abs(stats.zscore(outlier_res)) >= 3).any(axis=1)]
outlier_columns = list(outliers.T.columns)
outlier_columns.append('Class')
return df[outlier_columns]
"""
Proverava nedostajuce vrednosti
"""
def checkNaNvalues(df):
if df.isnull().values.any():
print("NaN vrednosti postoje\n")
else:
print("NaN vrednosti ne postoje\n")
def main():
print("PRETRPOCESIRANJE")
print("---------------------------------------------\n")
print("Ucitavanje podataka ... \n")
first_df = pd.read_csv('001_Melanoma_Cell_Line_csv.csv', index_col=0)
second_df = pd.read_csv('002_Melanoma_Cell_Line_csv.csv', index_col=0)
# Transponovanje matrice
first_df = first_df.T
second_df = second_df.T
print("Pocetna dimenzija 1. matrice: " + str(first_df.shape))
print("Pocetna dimenzija 2. matrice: " + str(second_df.shape) + '\n')
# Dodavanje klase
first_df.insert(len(first_df.columns), "Class", 'parental (BRAF inhibitor sensitive)')
second_df.insert(len(second_df.columns), "Class", 'BRAF inhibitor resistant')
print("Dimenzija 1. matrice posle dodavanja klase: " + str(first_df.shape))
print("Dimenzija 2. matrice posle dodavanja klase: " + str(second_df.shape) + '\n')
# Nalazenje kolona koje treba izbaciti na osnovu tipa (unija/presek)
columns_to_remove = zero_columns_to_remove(first_df, second_df, removal_type="intersection")
# Sjedinjavanje matrica
result_df = pd.concat([first_df, second_df], ignore_index=True)
print("Dimenzija spojene matrice: " + str(result_df.shape) + "\n")
# Provera nedostajucih vrednosti
checkNaNvalues(result_df)
# Eliminacija nula
result_df.drop(columns=columns_to_remove, inplace=True)
print("Dimenzija posle izbacenih nula: " + str(result_df.shape) + "\n")
# Pronalazenje autlajera
outliers_df = find_outlier_columns(result_df)
print("Dimenzija tabele autlajera: " + str(outliers_df.shape) + "\n")
# Random promesati podatke
result_df = result_df.sample(frac=1).reset_index(drop=True)
outliers_df = outliers_df.sample(frac=1).reset_index(drop=True)
# Uzimanje imena klasa za graf
columns_without_class = result_df.loc[:, result_df.columns != 'Class'].columns.tolist()
outlier_columns_without_class = outliers_df.loc[:, outliers_df.columns != 'Class'].columns.tolist()
# Razdvajanje test i trening podataka
x,y = seperate_data_class(result_df)
out_x, out_y = seperate_data_class(outliers_df)
x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.4)
ox_train, ox_test, oy_train, oy_test = train_test_split(out_x, out_y, test_size=0.4)
# TESTIRANJE STABLA ODLUCIVANJA - ceo skup
'''
decisionTree(x_train, x_test, y_train, y_test, 'gini', None)
decisionTree(x_train, x_test, y_train, y_test, 'gini', 6 )
decisionTree(x_train, x_test, y_train, y_test, 'gini', 4 )
decisionTree(x_train, x_test, y_train, y_test, 'gini', 3 , columns_without_class)
decisionTree(x_train, x_test, y_train, y_test, 'entropy', None)
decisionTree(x_train, x_test, y_train, y_test, 'gini', 6 )
decisionTree(x_train, x_test, y_train, y_test, 'entropy', 6, columns_without_class)
decisionTree(x_train, x_test, y_train, y_test, 'gini', 4, columns_without_class)
decisionTree(x_train, x_test, y_train, y_test, 'entropy', 4, columns_without_class)
'''
# TESTIRANJE KNN - ceo skup
'''
knn(x_train, x_test, y_train, y_test, 5, 'uniform', 'kd_tree')
knn(x_train, x_test, y_train, y_test, 5, 'uniform', 'ball_tree')
knn(x_train, x_test, y_train, y_test, 5, 'uniform')
knn(x_train, x_test, y_train, y_test, 5, 'uniform', 'brute')
knn(x_train, x_test, y_train, y_test, 3, 'uniform', 'brute')
knn(x_train, x_test, y_train, y_test, 5, 'uniform', 'brute')
knn(x_train, x_test, y_train, y_test, 10, 'uniform', 'brute')
knn(x_train, x_test, y_train, y_test, 15, 'uniform', 'brute')
'''
# TESTIRANJE SVM-a - ceo skup
'''
svm(x_train, x_test, y_train, y_test, 'rbf')
svm(x_train, x_test, y_train, y_test, 'poly')
svm(x_train, x_test, y_train, y_test, 'sigmoid')
'''
# TESTIRANJE BAGGING - ceo skup
'''
bagging(x_train, x_test, y_train, y_test, 3)
bagging(x_train, x_test, y_train, y_test, 5)
bagging(x_train, x_test, y_train, y_test, 7)
bagging(x_train, x_test, y_train, y_test, 10)
bagging(x_train, x_test, y_train, y_test, 5, 'svc')
bagging(x_train, x_test, y_train, y_test, 5, 'tree')
bagging(x_train, x_test, y_train, y_test, 5, 'knn')
'''
# TESTIRANJE BOOSTING - ceo skup
'''
boosting(x_train, x_test, y_train, y_test, 0.5)
boosting(x_train, x_test, y_train, y_test, 0.7)
boosting(x_train, x_test, y_train, y_test, 0.9)
boosting(x_train, x_test, y_train, y_test, 1)
boosting(x_train, x_test, y_train, y_test, 1.2)
boosting(x_train, x_test, y_train, y_test, 1, 'svc')
boosting(x_train, x_test, y_train, y_test, 1, 'tree')
'''
# TESTIRANJE redukovanog skupa
decisionTree(ox_train, ox_test, oy_train, oy_test, 'gini', 4, outlier_columns_without_class)
knn(ox_train, ox_test, oy_train, oy_test, 3, 'uniform', 'brute')
svm(ox_train, ox_test, oy_train, oy_test, 'poly')
bagging(ox_train, ox_test, oy_train, oy_test, 5, 'svc')
boosting(ox_train, ox_test, oy_train, oy_test, 1)
if __name__ == "__main__":
main()