python实现svm和使用f-score

发表于   |   分类于   |   阅读次数

使用方法

使用python语言实现对于支持向量机（SVM）特征选择的实现，特征选择算法为f-score,该程序的主要有点是可输入文件囊括了csv,libsvm,arff等在序列分类的机器学习领域常用到的格式，其中csv:最后一列为class,libsvm:第一列为class,arff:通常最后一列为类别，其中csv和libsvm中不存在开头，直接是使用的数据。

1
python train.py -i 1.csv,2.csv,3.libsvm,4.arff -c 5
  • 其中train.py为程序名称
  • -i :后面接文件名，可以为csv,libsvm,arff格式，多个文件也可以用，但建议不要，因为特征选择时间通常很长
  • -c:后面5代表五折交叉验证
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


#!/usr/bin/env python
# encoding:utf-8
import os
import sys
import getopt
import threading
import pandas as pd
import math
import numpy as np
from time import clock
from sklearn.feature_selection import  f_classif
from sklearn.externals.joblib import Memory
from sklearn import  metrics
import easy_excel

import itertools
from sklearn.model_selection import KFold  
from sklearn import svm
from sklearn.model_selection import train_test_split
import math
from xgboost import XGBClassifier
from sklearn.neighbors import KNeighborsClassifier
import easy_excel
from sklearn.preprocessing import MinMaxScaler
from sklearn.model_selection import *
import sklearn.ensemble
from sklearn.externals import joblib
from sklearn.linear_model import LogisticRegression
from sklearn import metrics
from sklearn.metrics import roc_curve, auc
import sys
from sklearn.model_selection import GridSearchCV
from sklearn.ensemble import RandomForestClassifier
from sklearn.naive_bayes import GaussianNB  
import subprocess
from sklearn.utils import shuffle
import itertools
from sklearn.ensemble import GradientBoostingClassifier
import sys
from sklearn.decomposition import PCA
from sklearn.model_selection import GridSearchCV
from sklearn.pipeline import Pipeline
from sklearn.svm import SVC, LinearSVC
from sklearn.naive_bayes import BernoulliNB
from sklearn.datasets import load_svmlight_file
from sklearn.tree import DecisionTreeClassifier
from sklearn.neighbors import KNeighborsClassifier
from sklearn.linear_model import SGDClassifier, LogisticRegression
from sklearn.ensemble import RandomForestClassifier, AdaBoostClassifier, \
    BaggingClassifier, ExtraTreesClassifier, GradientBoostingClassifier

def performance(labelArr, predictArr):
    #labelArr[i] is actual value,predictArr[i] is predict value
    TP = 0.; TN = 0.; FP = 0.; FN = 0.
    for i in range(len(labelArr)):
        if labelArr[i] == 1 and predictArr[i] == 1:
            TP += 1.
        if labelArr[i] == 1 and predictArr[i] == 0:
            FN += 1.
        if labelArr[i] == 0 and predictArr[i] == 1:
            FP += 1.
        if labelArr[i] == 0 and predictArr[i] == 0:
            TN += 1.
    if (TP + FN)==0:
        SN=0
    else:
        SN = TP/(TP + FN) #Sensitivity = TP/P  and P = TP + FN
    if (FP+TN)==0:
        SP=0
    else:
        SP = TN/(FP + TN) #Specificity = TN/N  and N = TN + FP
    if (TP+FP)==0:
        precision=0
    else:
        precision=TP/(TP+FP)
    if (TP+FN)==0:
        recall=0
    else:
        recall=TP/(TP+FN)
    GM=math.sqrt(recall*SP)
    #MCC = (TP*TN-FP*FN)/math.sqrt((TP+FP)*(TP+FN)*(TN+FP)*(TN+FN))
    return precision,recall,SN,SP,GM,TP,TN,FP,FN


mem = Memory("./mycache")
@mem.cache
def get_data(file_name):
    data = load_svmlight_file(file_name)
    return data[0], data[1]


def csv_and_arff2svm(arff_files):
    svm_files = []
    for arff_file in arff_files:
        name = arff_file[0: arff_file.rindex('.')]
        tpe = arff_file[arff_file.rindex('.')+1:]
        svm_file = name+".libsvm"
        svm_files.append(svm_file)
        if tpe == "arff":
            if os.path.exists(svm_file):
                pass
            else:
                f = open(arff_file)
                w = open(svm_file, 'w')
                flag = False
                for line in f.readlines():
                    if flag:
                        if line.strip() == '':
                            continue
                        temp = line.strip('\n').strip('\r').split(',')
                        w.write(temp[len(temp)-1])
                        for i in range(len(temp)-1):
                            w.write(' '+str(i+1)+':'+str(temp[i]))
                        w.write('\n')
                    else:
                        line = line.upper()
                        if line.startswith('@DATA') or flag:
                            flag = True
                f.close()
                w.close()
        elif tpe == "csv":
            if os.path.exists(svm_file):
                pass
            else:
                f = open(arff_file)
                w = open(svm_file, 'w')
                for line in f.readlines():
                    if line.strip() == '':
                        continue
                    temp = line.strip('\n').strip('\r').split(',')
                    w.write(temp[len(temp)-1])
                    for i in range(len(temp)-1):
                        w.write(' '+str(i+1)+':'+str(temp[i]))
                    w.write('\n')
                f.close()
                w.close()
        elif tpe == "libsvm":
            continue
        else:
            print "File format error! Arff and libsvm are passed."
            sys.exit()
    return svm_files
opts, args = getopt.getopt(sys.argv[1:], "hi:c:t:o:s:m:", )
for op, value in opts:
    if op == "-i":
        input_files = str(value)
        input_files = input_files.replace(" ", "").split(',')
        for input_file in input_files:
            if input_file == "":
                print "Warning: please insure no blank in your input files !"
                sys.exit()
    elif op == "-c":
        cv = int(value)
if __name__ =="__main__":
    path=""
    outputname="svm_f-score"
    name=outputname
    print '*** Validating file format ...'
    input_files = csv_and_arff2svm(input_files)
    for input_file in input_files:
        # 导入原始数据
        X, Y = get_data(input_file)
        train_data = X.todense()
        train_data=np.array(train_data)
        F, pval = f_classif(train_data, Y)
        idx = np.argsort(F)
        selected_list_=idx[::-1]
        F_sort_value=[F[e] for e in selected_list_]
        with open("all_dimension_results.txt",'a') as f:
                f.write(str(F_sort_value)+"\n")
        print F_sort_value
        with open("all_dimension_results.txt",'a') as f:
                f.write(str(selected_list_)+"\n")
        print selected_list_
        bestACC=0
        bestC=0
        bestgamma=0
        best_dimension=0
        all_dimension_results=[]
        for select_num in xrange(1,len(train_data[0])+1):
            train_data2=train_data
            print np.array(train_data).shape
            print np.array(train_data2).shape
            selected_list_2=selected_list_[xrange(select_num)]
            X_train=pd.DataFrame(train_data2)
            X_train=X_train.iloc[:,selected_list_2]
            X = np.array(X_train)
            svc = svm.SVC()
            parameters = {'kernel': ['rbf'], 'C':map(lambda x:2**x,np.linspace(-2,5,7)), 'gamma':map(lambda x:2**x,np.linspace(-5,2,7))}
            clf = GridSearchCV(svc, parameters, cv=cv, n_jobs=2, scoring='accuracy')
            clf.fit(X, Y)
            C=clf.best_params_['C']
            joblib.dump(clf,path+outputname+str(select_num)+".model")
            gamma=clf.best_params_['gamma']
            y_predict=cross_val_predict(svm.SVC(kernel='rbf',C=C,gamma=gamma),X,Y,cv=cv,n_jobs=2)
            y_predict_prob=cross_val_predict(svm.SVC(kernel='rbf',C=C,gamma=gamma,probability=True),X,Y,cv=cv,n_jobs=2,method='predict_proba')
            predict_save=[Y.astype(int),y_predict.astype(int),y_predict_prob[:,1]]
            predict_save=np.array(predict_save).T
            pd.DataFrame(predict_save).to_csv(path+outputname+"_"+'_predict_crossvalidation.csv',header=None,index=False)
            ROC_AUC_area=metrics.roc_auc_score(Y,y_predict)
            ACC=metrics.accuracy_score(Y,y_predict)
            precision, recall, SN, SP, GM, TP, TN, FP, FN = performance(Y, y_predict)
            F1_Score=metrics.f1_score(Y, y_predict)
            F_measure=F1_Score
            MCC=metrics.matthews_corrcoef(Y, y_predict)
            pos=TP+FN
            neg=FP+TN
            savedata=[[['svm'+"C:"+str(C)+"gamma:"+str(gamma),ACC,precision, recall,SN, SP, GM,F_measure,F1_Score,MCC,ROC_AUC_area,TP,FN,FP,TN,pos,neg]]]
            if ACC>bestACC:
                bestACC=ACC
                bestgamma=gamma
                bestC=C
                best_dimension=X.shape[1]
            print savedata
            print X.shape[1]
            with open("all_dimension_results.txt",'a') as f:
                f.write(str(savedata)+"\n")
            all_dimension_results.append(savedata)
        print bestACC
        print bestC
        print bestgamma
        print best_dimension
        easy_excel.save("svm_crossvalidation",[str(X.shape[1])],savedata,path+'cross_validation_'+name+'.xls')