Non_Gen_LinCFA/DdimensionalExperimentClass.py at main · climateintelligence/Non_Gen_LinCFA · GitHub

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
import sys
import pandas as pd
sys.path.append("../PaperNon_Gen_LinCFA/")
from NonLinCFA.NonLinCFA import NonLinCFA
from sklearn.linear_model import LogisticRegression,LinearRegression
from sklearn.metrics import r2_score,accuracy_score
import matplotlib.pyplot as plt
from NonLinCFA.NonLinCFA_anyFunction import NonLinCFA_anyFunction
from LinCFA.LinCFA import LinCFA
from GenLinCFA.GenLinCFA import GenLinCFA
from GenLinCFA.GenLinCFA_anyFunction import GenLinCFA_anyFunction
import numpy as np
from mlxtend.feature_selection import SequentialFeatureSelector as SFS
import random
from sklearn import preprocessing
import pickle
import argparse
from joblib import Parallel, delayed

def compute_accuracy(x_train, y_train, x_val, y_val):
    regr = LogisticRegression().fit(x_train,y_train)
    y_pred = regr.predict(x_val)
    return accuracy_score(y_val, y_pred)

def compute_wrapper(x_trainVal, y_trainVal,n_features):
    sfs = SFS(LogisticRegression(),
           k_features=n_features,
           forward=True,
           floating=False,
           scoring = 'accuracy',
           cv = 5,
	   n_jobs=5)

    sfs.fit(x_trainVal, y_trainVal)
    return pd.DataFrame(sfs.subsets_).transpose()

def squared_aggregation(x):
    x_sq = x**2
    return x_sq.sum(axis=1)

def mean_aggregation(x):
    return x.mean(axis=1)

def run_GenLinCFA(x,eps):
    output = GenLinCFA_anyFunction(x.iloc[:2000,:],'target', eps1=eps, n_val=5 , neigh=0, eps2=1, customFunction=mean_aggregation).compute_clusters()

    aggregate_x = pd.DataFrame()
    for i in range(len(output)):
        aggregate_x[i] = mean_aggregation(x[output[i]])

    actual_score = compute_accuracy(aggregate_x.iloc[:2000,:], x.loc[:,'target'].iloc[:2000], aggregate_x.iloc[2000:,:], x.loc[:,'target'].iloc[2000:])

    return [eps,len(output),actual_score]

def linear_experiment_noResampling(n_reps=10, n_variables=100, noise=10, hyper=[0.5]):
    NonLinCFA_score = []
    list_of_length = []
    wrapper_score = []
    LinCFA_score = []
    list_of_length_lin = []
    GenLinCFA_score = []
    list_of_length_gen = []

    x_all,y_all,coeffs = generate_dataset_n(n_data=3000*10, noise=noise, p1=0.3, p2=0.7, n_variables=n_variables, coeffs=[0])

    for trials in range(n_reps):
        x = pd.DataFrame(x_all[3000*trials:3000*(trials+1)])
        x['target'] = y_all[3000*trials:3000*(trials+1)]
        x['target'] = x.apply(lambda x: 1 if x.target>=0 else 0,axis=1)

        results = Parallel(n_jobs=10)(delayed(run_GenLinCFA)(x,eps) for eps in hyper)
        print(results)
        GenLinCFA_score.append(results)

        #for i in range(50):
        res = compute_wrapper(x.iloc[:2000,:-1], x.iloc[:2000,-1], 50)
        for i in range(50):
            x_wrapper_red = x[list(res.loc[i+1,'feature_names'])]
            wrapper_score.append(compute_accuracy(x_wrapper_red.iloc[:2000],x.loc[:,'target'].iloc[:2000],x_wrapper_red.iloc[2000:],x.loc[:,'target'].iloc[2000:]))
            print(i+1,wrapper_score[-1])

    return wrapper_score,GenLinCFA_score


### generate a dataset of n samples and standardize the variables x1,x2,x3
def generate_dataset_n(n_data=3000, noise=1, p1=0.5, p2=0.5, n_variables=10, coeffs=[0]):
    x = np.zeros((n_data,n_variables))

    rng = np.random.default_rng(seed=0)
    random.seed(30)

    x[:,0] = rng.uniform(0,1,n_data)
    if coeffs[0]==0:
        coeffs = rng.uniform(0,1,size=n_variables)

    for i in range(n_variables-1):
        j = random.randrange(i+1)
        x[:,i+1] = p1*rng.uniform(0,1,n_data) + p2*x[:,j]

    scaler = preprocessing.StandardScaler().fit(x)
    x = scaler.transform(x)

    delta = rng.normal(0, noise, size=(n_data,1))

    y = np.dot(x,coeffs).reshape(n_data,1) + delta.reshape(n_data,1)
    return x,y,coeffs

### generate a dataset of n samples with nonlinear relationship
def generate_dataset_quadratic(n_data=3000, noise=1, p1=0.5, p2=0.5, n_variables=10, coeffs=[0]):
    x = np.zeros((n_data,n_variables))

    rng = np.random.default_rng(seed=0)
    random.seed(30)

    x[:,0] = rng.uniform(0,1,n_data)
    if coeffs[0]==0:
        coeffs = rng.uniform(0,1,size=n_variables)

    for i in range(n_variables-1):
        j = random.randrange(i+1)
        x[:,i+1] = p1*rng.uniform(0,1,n_data) + p2*x[:,j]

    scaler = preprocessing.StandardScaler().fit(x)
    x = scaler.transform(x)

    delta = rng.normal(0, noise, size=(n_data,1))

    y = np.dot(x**2,coeffs).reshape(n_data,1) + delta.reshape(n_data,1)
    return x,y,coeffs


### main run
if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument("--n_variables", default=100, type=int)
    parser.add_argument("--noise", default=10, type=float)
    parser.add_argument("--n_repetitions", default=500, type=int)
    parser.add_argument("--p1", default=0.3, type=float)
    parser.add_argument("--p2", default=0.7, type=float)
    parser.add_argument("--results_file", default='res.pkl')
    parser.add_argument("--hyperparameter", nargs="*", default=[0.5], type=float)

    args = parser.parse_args()
    print(args)

    ##################### experiment ########################

    wrapper_score,GenLinCFA_score = linear_experiment_noResampling(n_reps=args.n_repetitions, n_variables=args.n_variables, noise=args.noise, hyper = args.hyperparameter)

    ##################### save the results ########################

    with open(args.results_file, 'wb') as f:
        pickle.dump([wrapper_score,GenLinCFA_score], f)