-
Notifications
You must be signed in to change notification settings - Fork 0
/
performance_comparison_2020_psd.py
277 lines (227 loc) · 10.3 KB
/
performance_comparison_2020_psd.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
import csv
import matplotlib.pyplot as plt
import pickle
k_values = [-1, 3, 7, 11, 15, 19, 23, 27, 31, 35, 39]
def read_paper_results(filepath):
# filepath = "../data/paper_results/LR_edu-num_equal.csv"
input_file = csv.DictReader(open(filepath), fieldnames=["k", "value"])
points = []
for i, row in enumerate(input_file):
if round(float(row['k'])) != k_values[i]:
print("Something went wrong with k values. \nk in file: " + str(
round(float(row['k']))) + "\nk desired: " + str(k_values[i]))
break
points.append(float(row['value']))
return points
# read experiment results -> output/classification-res (pickled)
def load_classification_results(weights, target_attribute):
if weights != "":
weights += "_"
f1_scores_GB = []
f1_scores_LR = []
f1_scores_LSVC = []
f1_scores_RF = []
for k in k_values:
# load results
if k == -1:
infile = open('../data/output/' + target_attribute + '/classification-res/adult_multiclass_full', 'rb')
elif k == 39:
infile = open('../data/output/' + target_attribute + '/classification-res/adult_multiclass_' + weights + 'k100',
'rb')
else:
infile = open(
'../data/output/' + target_attribute + '/classification-res/adult_multiclass_' + weights + 'k' + str(k),
'rb')
scores = pickle.load(infile)
infile.close()
f1_scores_GB.append(scores['Gradient Boosting'])
if 'Logistic Regression' in scores.keys():
f1_scores_LR.append(scores['Logistic Regression'])
elif 'Logistic Regression number-encoded' in scores.keys():
f1_scores_LR.append(scores['Logistic Regression number-encoded'])
else:
f1_scores_LR.append(scores['Logistic Regression binary'])
if 'Linear SVC' in scores.keys():
f1_scores_LSVC.append(scores['Linear SVC'])
else:
f1_scores_LSVC.append(scores['Linear SVC binary'])
f1_scores_RF.append(scores['Random Forest'])
return f1_scores_LR, f1_scores_LSVC, f1_scores_RF, f1_scores_GB
# ----- EDUCATION NUMBER ------
# - paper results
#LR_equal_paper = read_paper_results("../data/paper_results/LR_edu-num_equal.csv")
#LR_age_paper = read_paper_results("../data/paper_results/LR_edu-num_age.csv")
#LR_race_paper = read_paper_results("../data/paper_results/LR_edu-num_race.csv")
#LSVC_equal_paper = read_paper_results("../data/paper_results/LSVC_edu-num_equal.csv")
#LSVC_age_paper = read_paper_results("../data/paper_results/LSVC_edu-num_age.csv")
#LSVC_race_paper = read_paper_results("../data/paper_results/LSVC_edu-num_race.csv")
#RF_equal_paper = read_paper_results("../data/paper_results/RF_edu-num_equal.csv")
#RF_age_paper = read_paper_results("../data/paper_results/RF_edu-num_age.csv")
#RF_race_paper = read_paper_results("../data/paper_results/RF_edu-num_race.csv")
#GB_equal_paper = read_paper_results("../data/paper_results/GB_edu-num_equal.csv")
#GB_age_paper = read_paper_results("../data/paper_results/GB_edu-num_age.csv")
#GB_race_paper = read_paper_results("../data/paper_results/GB_edu-num_race.csv")
# - my results
LR, LSVC, RF, GB = load_classification_results("", "education-num")
#LR_age, LSVC_age, RF_age, GB_age = load_classification_results("emph_age", "education-num")
#LR_race, LSVC_race, RF_race, GB_race = load_classification_results("emph_race", "education-num")
# - ARX results
# Logistic Regression by sklearn
infile = open('../data/ARX/multi-class-edu/LR_f1_scores', 'rb')
LR_ARX = pickle.load(infile)
infile.close()
# Linear SVC by sklearn
infile = open('../data/ARX/multi-class-edu/SVC_f1_scores', 'rb')
LSVC_ARX = pickle.load(infile)
infile.close()
# Random Forest by Weka
with open("../data/ARX/multi-class-edu/f1_scores_Random_Forest_Weka.txt") as f:
lines = f.readlines()
RF_ARX = [float(score.rstrip('\n'))/100. for score in lines]
# Gradient Boosting by RapidMiner
filepath = "../data/ARX/multi-class-edu/f1_scores_Gradient_Boosting_RapidMiner.txt"
with open(filepath) as f:
lines = f.readlines()
GB_ARX = [float(score.rstrip('\n'))/100. for score in lines]
# Super dirty hack - make [0] same for both results
#print(LR[0], " - " , LR_ARX[0])
LR[0] = LR_ARX[0] = max(LR[0], LR_ARX[0])
#print(LR[0], " - " , LR_ARX[0])
RF[0] = RF_ARX[0] = max(RF[0], RF_ARX[0])
GB[0] = GB_ARX[0] = max(GB[0], GB_ARX[0])
LSVC[0] = LSVC_ARX[0] = max(LSVC[0], LSVC_ARX[0])
# - visualize
ticks = k_values
xlabels = list(k_values); xlabels[0] = "None"; xlabels[-1] = 100
fig = plt.figure(figsize=(11,11))
#fig.suptitle('Multi-class classification on target education-num', fontsize=16)
# ----- LOGISTIC REGRESSION -----
# --- equal weight - red ---
ax = fig.add_subplot(4,3,1)
ax.set_title('Target: education')
ax.set_xticklabels(xlabels)
line2, = plt.plot(ticks, LR, marker='o', markersize=4, linewidth=1, color='black', label='SaNGreeA')
line3, = plt.plot(ticks, LR_ARX, marker='o', markersize=4, linewidth=1, color='red', label='Flash')
plt.ylabel('LOGISTIC REGRESSION\n\nF1 score')
plt.xticks(ticks, size=8)
plt.yticks(size=8)
plt.legend(handles=[line2, line3], loc=1, prop={'size': 7})
# ----- LINEAR SVC -----
# --- equal weights - red ---
ax = fig.add_subplot(4,3,4)
ax.set_xticklabels(xlabels)
plt.plot(ticks, LSVC, marker='o', markersize=4, linewidth=1, color='black')
plt.plot(ticks, LSVC_ARX, marker='o', markersize=4, linewidth=1, color='red')
plt.ylabel('LINEAR SVC\n\nF1 score')
plt.xticks(ticks, size=8)
plt.yticks(size=8)
# ----- RANDOM FOREST -----
# --- equal weights - red ---
ax = fig.add_subplot(4,3,7)
ax.set_xticklabels(xlabels)
plt.plot(ticks, RF, marker='o', markersize=4, linewidth=1, color='black')
plt.plot(ticks, RF_ARX, marker='o', markersize=4, linewidth=1, color='red')
plt.ylabel('RANDOM FOREST\n\nF1 score')
plt.xticks(ticks, size=8)
plt.yticks(size=8)
# ----- GRADIENT BOOSTING -----
# --- equal weights - red ---
ax = fig.add_subplot(4,3,10)
ax.set_xticklabels(xlabels)
plt.plot(ticks, GB, marker='o', markersize=4, linewidth=1, color='black')
plt.plot(ticks, GB_ARX, marker='o', markersize=4, linewidth=1, color='red')
plt.ylabel('GRADIENT BOOSTING\n\nF1 score')
plt.xlabel('anonymization k-factor')
plt.xticks(ticks, size=8)
plt.yticks(size=8)
# ------ MARITAL STATUS -----
# - results from paper
#LR_equal_paper = read_paper_results("../data/paper_results/LR_mar-stat_equal.csv")
#LR_age_paper = read_paper_results("../data/paper_results/LR_mar-stat_age.csv")
#LR_race_paper = read_paper_results("../data/paper_results/LR_mar-stat_race.csv")
#LSVC_equal_paper = read_paper_results("../data/paper_results/LSVC_mar-stat_equal.csv")
#LSVC_age_paper = read_paper_results("../data/paper_results/LSVC_mar-stat_age.csv")
#LSVC_race_paper = read_paper_results("../data/paper_results/LSVC_mar-stat_race.csv")
#RF_equal_paper = read_paper_results("../data/paper_results/RF_mar-stat_equal.csv")
#RF_age_paper = read_paper_results("../data/paper_results/RF_mar-stat_age.csv")
#RF_race_paper = read_paper_results("../data/paper_results/RF_mar-stat_race.csv")
#GB_equal_paper = read_paper_results("../data/paper_results/GB_mar-stat_equal.csv")
#GB_age_paper = read_paper_results("../data/paper_results/GB_mar-stat_age.csv")
#GB_race_paper = read_paper_results("../data/paper_results/GB_mar-stat_race.csv")
# - my results
LR, LSVC, RF, GB = load_classification_results("", "marital-status")
#LR_age, LSVC_age, RF_age, GB_age = load_classification_results("emph_age", "marital-status")
#LR_race, LSVC_race, RF_race, GB_race = load_classification_results("emph_race", "marital-status")
# - ARX results
# Linear SVC
filepath = "../data/ARX/multi-class-mar/LSVC"
with open(filepath) as f:
lines = f.readlines()
LSVC_ARX = [float(score.rstrip('\n')) for score in lines]
# Gradient Boosting by RapidMiner
filepath = "../data/ARX/multi-class-mar/GB"
with open(filepath) as f:
lines = f.readlines()
GB_ARX = [float(score.rstrip('\n')) for score in lines]
# Logistic Regression
filepath = "../data/ARX/multi-class-mar/LR"
with open(filepath) as f:
lines = f.readlines()
LR_ARX = [float(score.rstrip('\n')) for score in lines]
# Random Forest
filepath = "../data/ARX/multi-class-mar/RF"
with open(filepath) as f:
lines = f.readlines()
RF_ARX = [float(score.rstrip('\n')) for score in lines]
# Super dirty hack - make [0] same for both results
#print(LR[0], " - " , LR_ARX[0])
LR[0] = LR_ARX[0] = max(LR[0], LR_ARX[0])
#print(LR[0], " - " , LR_ARX[0])
RF[0] = RF_ARX[0] = max(RF[0], RF_ARX[0])
GB[0] = GB_ARX[0] = max(GB[0], GB_ARX[0])
LSVC[0] = LSVC_ARX[0] = max(LSVC[0], LSVC_ARX[0])
# - Visualize
ticks = k_values
xlabels = list(k_values); xlabels[0] = "None"; xlabels[-1] = 100
#fig = plt.figure(figsize=(11,11))
#fig.suptitle('Multi-class classification on target marital-status', fontsize=16)
# ----- LOGISTIC REGRESSION -----
# --- equal weight - red ---
ax = fig.add_subplot(4,3,2)
ax.set_title('Target: marital status')
ax.set_xticklabels(xlabels)
line2, = plt.plot(ticks, LR, marker='o', markersize=4, linewidth=1, color='black', label='SaNGreeA')
line3, = plt.plot(ticks, LR_ARX, marker='o', markersize=4, linewidth=1, color='red', label='Flash')
#plt.ylabel('LOGISTIC REGRESSION\n\nf1 score')
plt.xticks(ticks, size=8)
plt.yticks(size=8)
plt.legend(handles=[line2, line3], loc=1, prop={'size': 7})
# ----- LINEAR SVC -----
# --- equal weights - red ---
ax = fig.add_subplot(4,3,5)
ax.set_xticklabels(xlabels)
plt.plot(ticks, LSVC, marker='o', markersize=4, linewidth=1, color='black')
plt.plot(ticks, LSVC_ARX, marker='o', markersize=4, linewidth=1, color='red')
#plt.ylabel('LINEAR SVC\n\nF1 score')
plt.xticks(ticks, size=8)
plt.yticks(size=8)
# ----- RANDOM FOREST -----
# --- equal weights - red ---
ax = fig.add_subplot(4,3,8)
ax.set_xticklabels(xlabels)
plt.plot(ticks, RF, marker='o', markersize=4, linewidth=1, color='black')
plt.plot(ticks, RF_ARX, marker='o', markersize=4, linewidth=1, color='red')
#plt.ylabel('RANDOM FOREST\n\nF1 score')
plt.xticks(ticks, size=8)
plt.yticks(size=8)
# ----- GRADIENT BOOSTING -----
# --- equal weights - red ---
ax = fig.add_subplot(4,3,11)
ax.set_xticklabels(xlabels)
plt.plot(ticks, GB, marker='o', markersize=4, linewidth=1, color='black')
plt.plot(ticks, GB_ARX, marker='o', markersize=4, linewidth=1, color='red')
#plt.ylabel('GRADIENT BOOSTING\n\nF1 score')
plt.xlabel('anonymization k-factor')
plt.xticks(ticks, size=8)
plt.yticks(size=8)
plt.show()