-
Notifications
You must be signed in to change notification settings - Fork 0
/
utils_recourse.py
162 lines (136 loc) · 5.05 KB
/
utils_recourse.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
from sklearn.cluster import KMeans
from sklearn.metrics import pairwise_distances
import numpy as np
import pandas as pd
import seaborn as sns
def get_reward(ml_model, true_cf_received, rewardtype = 'validation'):
# compute the reward for each user
labels_predicted = ml_model.predict(true_cf_received)
if rewardtype == 'validation':
reward = sum(labels_predicted>0.5)
return reward
def set_style():
# This sets reasonable defaults for font size for
# a figure that will go in a paper
sns.set_context("paper")
# Set the font to be serif, rather than sans
sns.set(font='serif')
# Make the background white, and specify the
# specific font family
sns.set_style("white", {
"font.family": "serif",
"font.serif": ["Times New Roman", "Palatino", "serif"]
})
def seed_everything(seed: int):
import random, os
import numpy as np
import torch
random.seed(seed)
os.environ['PYTHONHASHSEED'] = str(seed)
np.random.seed(seed)
torch.manual_seed(seed)
torch.cuda.manual_seed(seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = True
def encode_constraint(cf, negative, sens_name):
for sen_name in sens_name:
cf[sen_name] = negative[sen_name]
return cf
def invalidation(cf, ml_model):
inv = ml_model.predict(cf)
return np.mean(inv < 0.5)
def centralization(sens, features, radius = None):
if sum(sens==0) > sum(sens==1):
sensid = 1
else:
sensid = 0
minority = features[sens==sensid]
majority = features[sens==1-sensid]
distance_cross = pairwise_distances(minority, majority, metric='euclidean')
distance_in = pairwise_distances(minority, minority, metric='euclidean')
if radius is None:
radius = np.quantile(distance_in, 0.1) # syn 0.1
central_region = ((np.sum(distance_in < radius, 1)) / ((np.sum(distance_cross < radius, 1))+(np.sum(distance_in < radius, 1))))
central_region = central_region > 0.5
num_central = sum([i==sensid and j ==1 for i,j in zip(sens, central_region)]) * 1.
ci = num_central / sum(sens == sensid)
return ci, radius
def avg_proximity2(sens, features, origin_features = None):
if sum(sens==0) > sum(sens==1):
sensid = 1
else:
sensid = 0
C_ij = pairwise_distances(features, metric='l2')
C_ij = np.exp(-C_ij)
a = 0
m0 = sum(sens == sensid)
m1 = sum(sens == 1-sensid)
num_classes = C_ij.shape[0]
a = np.sum((C_ij.T * np.array(sens==sensid)).T * np.array(1-(sens==sensid))) / (m0 * m1)
return a
def atkinson(sens, features, origin_features = None, beta = 0.5):
# how to define a neighborhood
from sklearn.cluster import KMeans
from sklearn.metrics import pairwise_distances
if sum(sens==0) > sum(sens==1):
sensid = 1
else:
sensid = 0
nn = 30
cluster = KMeans(n_clusters = nn, random_state = 0)
if origin_features is not None:
cluster.fit(origin_features)
neighborhood = cluster.predict(features)
num_classes = nn
P = sum(sens==sensid) / len(sens)
T = len(sens)
ak = 0
for i in range(num_classes):
ti = sum(neighborhood==i)
if ti == 0:
continue
mi = sum(sens[neighborhood==i] == sensid)
pi = mi / ti
aki = (1-pi) ** (1-beta) * (pi ** beta) * ti / (T * P)
ak += aki
ak = 1 - P/(1-P) * np.abs(ak)**(1/(1-beta))
return ak
def recourse_cost(origin_features, new_features):
return np.sqrt(((origin_features - new_features)**2).sum(1)).mean()
def fairness_cost(origin_features, new_features, sens):
rec1 = np.sqrt(((origin_features[sens==1] - new_features[sens==1])**2).sum(1)).mean()
rec2 = np.sqrt(((origin_features[sens==0] - new_features[sens==0])**2).sum(1)).mean()
print(rec1)
print(rec2)
print(np.abs(rec1-rec2))
print(origin_features.shape)
print(sens.shape)
return np.abs(rec1-rec2)
import numpy as np
import pandas as pd
from sklearn.neighbors import NearestNeighbors
from carla.evaluation import remove_nans
from carla.evaluation.api import Evaluation
def ynn(factuals, cf, ml_model, num = 5):
factuals = ml_model.get_ordered_features(ml_model.data.df)
cf = cf[factuals.columns]
cf = cf.fillna(cf.mean(skipna=False))
cf = cf.fillna(0)
number_of_diff_labels = 0
nbrs = NearestNeighbors(n_neighbors=num).fit(factuals.values)
for i, row in cf.iterrows():
knn = nbrs.kneighbors(
row.values.reshape((1, -1)), num, return_distance=False
)[0]
for idx in knn:
neighbour = factuals.iloc[idx]
neighbour = neighbour.values.reshape((1, -1))
neighbour_label = np.argmax(ml_model.predict_proba(neighbour))
number_of_diff_labels += np.abs(1 - neighbour_label)
return 1 - (1 / (len(cf) * num)) * number_of_diff_labels
def closeness(cf, positive_points):
cf = cf.fillna(cf.mean(skipna=False))
cf = cf.fillna(0)
distance = pairwise_distances(cf, positive_points)
distance = distance.mean()
return distance