-
Notifications
You must be signed in to change notification settings - Fork 0
/
greedy_alg.py
136 lines (112 loc) · 3.98 KB
/
greedy_alg.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
import numpy as np
import pandas as pd
import sklearn as sk
from sklearn.preprocessing import PolynomialFeatures
from sklearn.linear_model import LinearRegression
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import r2_score
from statsmodels.api import OLS
import itertools
import math
import random
import scipy.sparse
from sklearn.externals.joblib.parallel import Parallel, delayed
import boto3, botocore
import time
import warnings
import multiprocessing
import scipy
from sklearn import preprocessing
warnings.filterwarnings("ignore")
# params
# can choose bayes_synth, linear_synth or log_synth for dat
dat = 'linear_synth'
print(dat)
k = 10
isParallel = False
data_folder = 'data/'
num_cores = multiprocessing.cpu_count()
print('num_cores:')
print(num_cores)
print('Fetching Files')
if dat == 'linear_synth':
X = np.load(data_folder + 'synth_lin_X.npy')
y_cat = np.load(data_folder + 'synth_lin_y.npy')
all_predictors = list(range(X.shape[1]))
elif dat == 'log_synth':
X = np.load(data_folder + 'synth_log_X.npy')
y_cat = np.load(data_folder + 'synth_log_y.npy')
all_predictors = list(range(X.shape[1]))
elif dat == 'bayes_synth':
X = np.load(data_folder + 'synth_bayes_X.npy')
all_predictors = list(range(X.shape[1]))
X = preprocessing.normalize(X, norm='l2', axis=0)
d = X.shape[0]
Lambda = np.eye(d)
sigma = 1
print('Objective function')
if 'linear' in dat:
def obj_fun(x_t, y_t):
model = OLS(y_t, x_t).fit()
pred = model.predict(x_t)
return r2_score(y_t, pred)
elif 'bayes' in dat:
def obj_fun(S):
trace_sig = np.trace(np.linalg.inv(Lambda))
X1 = X[:,S]
return trace_sig - np.trace(np.linalg.inv(Lambda + sigma**(-2)*np.matmul(X1, X1.T)))
elif 'log' in dat:
def log_likelihood(features, target, weights):
scores = np.dot(features, weights)
ll = np.sum( target*scores - np.log(1 + np.exp(scores)) )
return ll
# we use log likelihood to maximize but plot classification rate
def obj_fun(x_t, y_t):
logitm = LogisticRegression(C = 1e15,solver='lbfgs')
logitm.fit (x_t, y_t)
y_logit = logitm.predict(x_t)
class_rate = np.sum(y_logit == y_t)/len(y_t)
return 1/(-1*log_likelihood(x_t, y_t, logitm.coef_[0]))
def get_class_rate(x_t, y_t):
# Create logistic regression object
logitm = LogisticRegression(C = 1e15,solver='lbfgs')
logitm.fit (x_t, y_t)
y_logit = logitm.predict(x_t)
class_rate = np.sum(y_logit == y_t)/len(y_t)
return class_rate
def evaluate_oracle(best_k_predictors, a):
k_plus_1 = list(best_k_predictors + [a])
if 'bayes' in dat:
res = obj_fun(k_plus_1)
else:
x_train = X[:,k_plus_1]
res = get_class_rate(x_train, y_cat)
return res
def run_greedy(kToSelect, parallel=True):
# get starting set of data
predictors = [([], 0, 0)]
# loop through predictors and at each step,
# add one predictor that increases obj function the most
# and calculate obj function
for k in range(kToSelect):
best_k_predictors = predictors[-1][0]
r2 = []
predictor_list = list(set(all_predictors) - set(best_k_predictors))
if not parallel:
for predictor in predictor_list:
k_plus_1 = list(best_k_predictors + [predictor])
if 'bayes' in dat:
r2.append(obj_fun(k_plus_1))
else:
x_train = X[:,k_plus_1]
r2.append(obj_fun(x_train, y_cat))
else:
r2 = Parallel(n_jobs=-1)(delayed(evaluate_oracle)(best_k_predictors, a) for a in predictor_list)
best_k_plus_1 = best_k_predictors + [predictor_list[np.argmax(r2)]]
end = time.time()
timePassed = end-start
predictors.append((best_k_plus_1, np.max(r2), timePassed))
return predictors
start = time.time()
predictors = run_greedy(k, isParallel)
print(predictors)