forked from jrubin01/grasp-and-lift
-
Notifications
You must be signed in to change notification settings - Fork 0
/
simple_grasp_cross-validation.py
146 lines (121 loc) · 4.08 KB
/
simple_grasp_cross-validation.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
# -*- coding: utf-8 -*-
"""
Created on Tue Jul 7 18:40:37 CEST 2015
@author: Elena Cuoco
simple starting script, without the use of MNE
Thanks to @author: alexandrebarachant for his wornderful starting script
"""
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from sklearn.linear_model import LogisticRegression
from glob import glob
import os
from sklearn.preprocessing import StandardScaler
from sklearn.cross_validation import LeaveOneLabelOut
from sklearn.metrics import roc_auc_score
from joblib import Parallel, delayed
#############function to read data###########
def prepare_data_train(fname):
""" read and prepare training data """
# Read data
data = pd.read_csv(fname)
# events file
events_fname = fname.replace('_data','_events')
# read event file
labels= pd.read_csv(events_fname)
clean=data.drop(['id' ], axis=1)#remove id
labels=labels.drop(['id' ], axis=1)#remove id
return clean,labels
def prepare_data_test(fname):
""" read and prepare test data """
# Read data
data = pd.read_csv(fname)
return data
scaler= StandardScaler()
def data_preprocess_train(X):
X_prep=scaler.fit_transform(X)
#do here your preprocessing
return X_prep
def data_preprocess_test(X):
X_prep=scaler.transform(X)
#do here your preprocessing
return X_prep
def fit(X,y):
# Do here you training
clf = LogisticRegression()
clf.fit(X,y)
return clf
def predict(clf,X):
# do here your prediction
preds = clf.predict_proba(X)
return np.atleast_2d(preds[:,clf.classes_==1])
# training subsample.if you want to downsample the training data
subsample = 100
#series used for CV
series = range(2,9)
#######columns name for labels#############
cols = ['HandStart','FirstDigitTouch',
'BothStartLoadPhase','LiftOff',
'Replace','BothReleased']
#######number of subjects###############
subjects = range(1,13)
auc_tot = []
pred_tot = []
y_tot = []
###loop on subjects and 8 series for train data + 2 series for test data
for subject in subjects:
y_raw= []
raw = []
sequence = []
################ READ DATA ################################################
for ser in series:
fname = '../input/train/subj%d_series%d_data.csv' % (subject,ser)
data,labels=prepare_data_train(fname)
raw.append(data)
y_raw.append(labels)
sequence.extend([ser]*len(data))
X = pd.concat(raw)
y = pd.concat(y_raw)
#transform in numpy array
#transform train data in numpy array
X = np.asarray(X.astype(float))
y = np.asarray(y.astype(float))
sequence = np.asarray(sequence)
################ Train classifiers ########################################
cv = LeaveOneLabelOut(sequence)
pred = np.empty((X.shape[0],6))
for train, test in cv:
X_train = X[train]
X_test = X[test]
y_train = y[train]
#apply preprocessing
X_train=data_preprocess_train(X_train)
X_test=data_preprocess_test(X_test)
clfs = Parallel(n_jobs=6)(delayed(fit)(X_train[::subsample,:],y_train[::subsample,i]) for i in range(6))
preds = Parallel(n_jobs=6)(delayed(predict)(clfs[i],X_test) for i in range(6))
pred[test,:] = np.concatenate(preds,axis=1)
pred_tot.append(pred)
y_tot.append(y)
# get AUC
auc = [roc_auc_score(y[:,i],pred[:,i]) for i in range(6)]
auc_tot.append(auc)
print(auc)
pred_tot = np.concatenate(pred_tot)
y_tot = np.concatenate(y_tot)
global_auc = [roc_auc_score(y_tot[:,i],pred_tot[:,i]) for i in range(6)]
print('Global AUC : %.4f' % np.mean(global_auc))
auc_tot = np.asarray(auc_tot)
results = pd.DataFrame(data=auc_tot, columns=cols, index=subjects)
results.to_csv('results_cv_auc.csv')
plt.figure(figsize=(4,3))
results.mean(axis=1).plot(kind='bar')
plt.xlabel('Subject')
plt.ylabel('AUC')
plt.title('CV auc for each subject')
plt.savefig('cross_val_auc_subject.png' ,bbox_inches='tight')
plt.figure(figsize=(4,3))
results.mean(axis=0).plot(kind='bar')
plt.ylabel('AUC')
plt.title('CV auc for each class')
plt.savefig('cross_val_auc_class.png' ,bbox_inches='tight')