/
MyDecisionTreeClassifier.py
286 lines (251 loc) · 16.9 KB
/
MyDecisionTreeClassifier.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
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.metrics import f1_score
from sklearn.tree import DecisionTreeClassifier
import numpy as np
from sklearn.ensemble import AdaBoostClassifier,RandomForestClassifier
import glob
import pandas as pd
from sklearn.linear_model import LogisticRegression
import os
from sklearn.utils import shuffle
import collections
from sklearn.neural_network import MLPClassifier
from collections import OrderedDict
import matplotlib.pyplot as plt
def formatString(stringTemplate, *args, **kwargs):
# Replace any positional parameters
for i in range(0, len(args)):
tmp = '{%s}' % str(1 + i)
while True:
pos = stringTemplate.find(tmp)
if pos < 0:
break
stringTemplate = stringTemplate[:pos] + \
str(args[i]) + \
stringTemplate[pos + len(tmp):]
# Replace any named parameters
for key, val in kwargs.items():
tmp = '{%s}' % key
while True:
pos = stringTemplate.find(tmp)
if pos < 0:
break
stringTemplate = stringTemplate[:pos] + \
str(val) + \
stringTemplate[pos + len(tmp):]
return stringTemplate
class MyClassifiers():
'Handles Predicting Phishing URL by implementing scikit-learn DecisionTree Classifier'
def classify(self,realTrainFeatures, realTrainLabels, realTest_Features, realTest_Labels):
dir_name = os.path.dirname(os.path.realpath(__file__))
#predictionResult = open(dir_name+'/'+technique+'DecisionTreeResultsNoOversampling.txt','a+')
#predictionResult.truncate()
path = dir_name+'/LACityFake/*'
plt.grid(True)
figCount = 1
for dirName in glob.glob(path):
plt.clf()
fileName = dirName + '/scaled_fake_tabular.pickle'
print(dirName)
fakeDf = pd.read_pickle(fileName)
#print(fakeDf.columns)
#Generate labels.....
labels = []
for i in range(len(fakeDf)):
if fakeDf[i][8] > 77636.3656547:
labels.append(1) # rich
else:
labels.append(0) # poor
print(collections.Counter(labels))
algorithms = ['DecisionTree','RandomForest','AdaBoost','MLPClassifier']
fakeDfUpdated = np.delete(fakeDf, [0,1,2,3,4,5,6,7,8,9,11,15], axis=1)
# fakeDf.drop('Permanent Bonus Pay', axis=1, inplace=True)
# fakeDf.drop('Other Pay (Payroll Explorer)', axis=1, inplace=True)
# fakeDf.drop('Temporary Bonus Pay', axis=1, inplace=True)
# fakeDf.drop('% Over Base Pay', axis=1, inplace=True)
# fakeDf.drop('Q4 Payments', axis=1, inplace=True)
# fakeDf.drop('Q3 Payments', axis=1, inplace=True)
# fakeDf.drop('Q2 Payments', axis=1, inplace=True)
# fakeDf.drop('Q1 Payments', axis=1, inplace=True)
# fakeDf.drop('Projected Annual Salary', axis=1, inplace=True)
# fakeDf.drop('Total Payments', axis=1, inplace=True)
# fakeDf.drop('Payments Over Base Pay', axis=1, inplace=True)
# fakeDf.drop('Base Pay', axis=1, inplace=True)
# Create a dual for loop for everyalgorithm
for eachAlgorithm in algorithms:
if eachAlgorithm == 'DecisionTree':
parameters_decisionTree = {0: "criterion=gini,splitter=best,max_depth=None,min_samples_split=2,min_samples_leaf=1,max_features=None,presort=False",
1: "criterion=entropy,splitter=random,max_depth=None,min_samples_split=3,min_samples_leaf=1,max_features=None,presort=False",
2: "criterion=gini,splitter=best,max_depth=None,min_samples_split=2,min_samples_leaf=1,max_features=None,presort=False",
3: "criterion=entropy,splitter=random,max_depth=None,min_samples_split=4,min_samples_leaf=1,max_features=None,presort=False",
4: "criterion=gini,splitter=best,max_depth=None,min_samples_split=5,min_samples_leaf=1,max_features=None,presort=True",
5: "criterion=gini,splitter=best,max_depth=None,min_samples_split=6,min_samples_leaf=1,max_features=None,presort=False",
6: "criterion=gini,splitter=best,max_depth=None,min_samples_split=7,min_samples_leaf=1,max_features=None,presort=False",
7: "criterion=gini,splitter=best,max_depth=None,min_samples_split=8,min_samples_leaf=1,max_features=None,presort=True",
8: "criterion=gini,splitter=best,max_depth=None,min_samples_split=9,min_samples_leaf=1,max_features=None,presort=False",
9: "criterion=gini,splitter=best,max_depth=None,min_samples_split=10,min_samples_leaf=1,max_features=None,presort=True"
}
for i in range(0,10):
parameter = list(parameters_decisionTree.values())[i]
param = parameter.split(',')
my_criterion = param[0].split('=')[1]
my_splitter = param[1].split('=')[1]
my_max_depth = param[2].split('=')[1]
my_min_samples_split = param[3].split('=')[1]
my_min_samples_leaf = param[4].split('=')[1]
my_max_features = param[5].split('=')[1]
my_presort = param[6].split('=')[1]
#Every i represents one parameter setup..Define the parameters
clf = DecisionTreeClassifier(criterion=my_criterion,splitter=my_splitter,max_depth=None,min_samples_split=int(my_min_samples_split),
min_samples_leaf=int(my_min_samples_leaf),presort=my_presort)
clf.fit(realTrainFeatures,realTrainLabels)
realResult = clf.predict(realTest_Features)
f1score = f1_score(realTest_Labels,realResult,average='macro')
print("The f1_score is:" + str(f1score))
# Process Every Fake file....training on Fake and testing on Real
#Index to generate labels is Total Payments @index 8
#fakeDfUpdated = np.delete(fakeDf, [0,1,2,3,4,5,6,7,8,9,11,15], axis=1)
# ----------------
#-----------------
clf.fit(fakeDfUpdated,labels)
fakeResults = clf.predict(realTest_Features)
fakeF1 = f1_score(realTest_Labels,fakeResults,average='macro')
print('Fake Results F1-measure is: '+str(fakeF1))
# prepare real values...replicate
plt.title(dirName[57:73])
plt.plot([0, 1], [0, 1])
dt = plt.plot(f1score,fakeF1,'ro',label='Decision Tree')
plt.grid(True)
plt.xticks([0, 0.10, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, 1.0])
plt.yticks([0, 0.10, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, 1.0])
if eachAlgorithm == 'AdaBoost':
parameters_AdaBoost = {
0: "n_estimators=50,learning_rate=1,algorithm=SAMME",
1: "n_estimators=80,learning_rate=1.4,algorithm=SAMME.R",
2: "n_estimators=100,learning_rate=1.3,algorithm=SAMME",
3: "n_estimators=120,learning_rate=1.6,algorithm=SAMME.R",
4: "n_estimators=250,learning_rate=2.0,algorithm=SAMME",
5: "n_estimators=220,learning_rate=2.3,algorithm=SAMME.R",
6: "n_estimators=176,learning_rate=2.2,algorithm=SAMME",
7: "n_estimators=300,learning_rate=3.0,algorithm=SAMME.R",
8: "n_estimators=169,learning_rate=1.2,algorithm=SAMME",
9: "n_estimators=211,learning_rate=2.6,algorithm=SAMME.R"
}
for i in range(0,10):
f1score = 0
fakeF1 = 0
parameter = list(parameters_AdaBoost.values())[i]
param = parameter.split(',')
my_n_estimators = int(param[0].split('=')[1])
my_learning_rate = float(param[1].split('=')[1])
my_algorithm = param[2].split('=')[1]
#Every i represents one parameter setup..Define the parameters
clf = AdaBoostClassifier(n_estimators=my_n_estimators,learning_rate=my_learning_rate,algorithm=my_algorithm)
clf.fit(realTrainFeatures,realTrainLabels)
realResult = clf.predict(realTest_Features)
f1score = f1_score(realTest_Labels,realResult,average='macro')
print("The f1_score is:" + str(f1score))
clf.fit(fakeDfUpdated, labels)
fakeResults = clf.predict(realTest_Features)
fakeF1 = f1_score(realTest_Labels, fakeResults, average='macro')
print('Fake Results F1-measure is: ' + str(fakeF1))
# prepare real values...replicate
plt.plot([0, 1], [0, 1])
ab = plt.plot(f1score, fakeF1, 'bo',label='Ada Boost')
plt.grid(True)
plt.xticks([0, 0.10, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, 1.0])
plt.yticks([0, 0.10, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, 1.0])
if eachAlgorithm == 'RandomForest':
parameters_RandomForest = {
0: "n_estimators=10, criterion=gini, min_samples_split=2,max_features=auto,bootstrap=True,verbose=0",
1: "n_estimators=20, criterion=entropy, min_samples_split=3,max_features=auto,bootstrap=False,verbose=2",
2: "n_estimators=30, criterion=gini, min_samples_split=2,max_features=auto,bootstrap=True,verbose=4",
3: "n_estimators=40, criterion=entropy, min_samples_split=4,max_features=auto,bootstrap=False,verbose=6",
4: "n_estimators=50, criterion=gini, min_samples_split=5,max_features=auto,bootstrap=True,verbose=1",
5: "n_estimators=60, criterion=entropy, min_samples_split=6,max_features=auto,bootstrap=False,verbose=3",
6: "n_estimators=70, criterion=gini, min_samples_split=7,max_features=auto,bootstrap=True,verbose=1",
7: "n_estimators=80, criterion=entropy, min_samples_split=7,max_features=auto,bootstrap=False,verbose=5",
8: "n_estimators=90, criterion=gini, min_samples_split=8,max_features=auto,bootstrap=True,verbose=1",
9: "n_estimators=100, criterion=entropy, min_samples_split=10,max_features=auto,bootstrap=False,verbose=7"
}
for i in range(0,9):
f1score = 0
fakeF1 = 0
parameter = list(parameters_RandomForest.values())[i]
param = parameter.split(',')
RF_n_estimators = int(param[0].split('=')[1])
RF_criterion = param[1].split('=')[1]
RF_min_samples_split = int(param[2].split('=')[1])
RF_max_features= param[3].split('=')[1]
RF_bootstrap= param[4].split('=')[1]
RF_verbose= int(param[5].split('=')[1])
#Every i represents one parameter setup..Define the parameters
clf = RandomForestClassifier(n_estimators=RF_n_estimators,criterion=RF_criterion,min_samples_split=RF_min_samples_split,
max_features=RF_max_features,bootstrap=RF_bootstrap,verbose=RF_verbose)
clf.fit(realTrainFeatures,realTrainLabels)
realResult = clf.predict(realTest_Features)
f1score = f1_score(realTest_Labels,realResult,average='macro')
print("The f1_score is:" + str(f1score))
#fakeDfUpdated = np.delete(fakeDf, [6, 8, 9], axis=1)
clf.fit(fakeDfUpdated, labels)
fakeResults = clf.predict(realTest_Features)
fakeF1 = f1_score(realTest_Labels, fakeResults, average='macro')
print('Fake Results F1-measure is: ' + str(fakeF1))
# prepare real values...replicate
plt.plot([0, 1], [0, 1])
rf = plt.plot(f1score, fakeF1, 'go',label='Random Forest')
plt.grid(True)
plt.xticks([0,0.10, 0.20,0.30, 0.40,0.50, 0.60, 0.70, 0.80, 0.90, 1.0])
plt.yticks([0,0.10, 0.20,0.30, 0.40,0.50, 0.60, 0.70, 0.80, 0.90, 1.0])
if eachAlgorithm == 'MLPClassifier':
parameters_MLP = {
0: "activation=relu, solver=adam, learning_rate=constant,max_iter=200",
1: "activation=identity, solver=lbfgs, learning_rate=invscaling,max_iter=300",
2: "activation=logistic, solver=sgd, learning_rate=adaptive,max_iter=250",
3: "activation=tanh, solver=adam, learning_rate=constant,max_iter=320",
4: "activation=relu, solver=lbfgs, learning_rate=constant,max_iter=270",
5: "activation=identity, solver=lbfgs, learning_rate=invscaling,max_iter=270",
6: "activation=logistic, solver=lbfgs, learning_rate=constant,max_iter=350",
7: "activation=tanh, solver=lbfgs, learning_rate=constant,max_iter=400",
8: "activation=relu, solver=lbfgs, learning_rate=constant,max_iter=325",
9: "activation=logistic, solver=sgd, learning_rate=constant,max_iter=480"
}
for i in range(0,10):
f1score = 0
fakeF1 = 0
parameter = list(parameters_MLP.values())[i]
param = parameter.split(',')
MLP_activation = param[0].split('=')[1]
MLP_solver = param[1].split('=')[1]
MLP_learningRate = param[2].split('=')[1]
MLP_max_iter = int(param[3].split('=')[1])
# Every i represents one parameter setup..Define the parameters
clf = MLPClassifier(activation=MLP_activation,solver=MLP_solver,learning_rate=MLP_learningRate,
max_iter=MLP_max_iter)
clf.fit(realTrainFeatures, realTrainLabels)
print('PRINTING REAL...')
print(list(realTrainLabels).count(0.0))
print(list(realTrainLabels).count(1.0))
realResult = clf.predict(realTest_Features)
f1score = f1_score(realTest_Labels, realResult, average='macro')
print("The f1_score is:" + str(f1score))
# fakeDfUpdated = np.delete(fakeDf, [6, 8, 9], axis=1)
print('PRINTING FROM FAKE')
print(labels.count(0.0))
print(labels.count(1.0))
clf.fit(fakeDfUpdated, labels)
fakeResults = clf.predict(realTest_Features)
fakeF1 = f1_score(realTest_Labels, fakeResults, average='macro')
print('Fake Results F1-measure is: ' + str(fakeF1))
# prepare real values...replicate
plt.plot([0, 1], [0, 1])
lr = plt.plot(f1score, fakeF1, 'co', label='MLP Classifier')
plt.grid(True)
plt.xticks([0, 0.10, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, 1.0])
plt.yticks([0, 0.10, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, 1.0])
handles, labels2 = plt.gca().get_legend_handles_labels()
by_label = OrderedDict(zip(labels2, handles))
plt.legend(by_label.values(), by_label.keys(),loc='best')
plt.savefig(str(figCount)+'.png',format='png',dpi=300)
figCount+=1
plt.clf()