python svm 实例_Python svm.predict方法代码示例

# 需要导入模块: from sklearn import svm [as 别名]

# 或者: from sklearn.svm import predict [as 别名]

def classifier_wrapper(classifier, classifier_type, test_sample):

"""

This function is used as a wrapper to pattern classification.

ARGUMENTS:

- classifier: a classifier object of type sklearn.svm.SVC or

kNN (defined in this library) or sklearn.ensemble.

RandomForestClassifier or sklearn.ensemble.

GradientBoostingClassifier or

sklearn.ensemble.ExtraTreesClassifier

- classifier_type: "svm" or "knn" or "randomforests" or

"gradientboosting" or "extratrees"

- test_sample: a feature vector (np array)

RETURNS:

- R: class ID

- P: probability estimate

EXAMPLE (for some audio signal stored in array x):

import audioFeatureExtraction as aF

import audioTrainTest as aT

# load the classifier (here SVM, for kNN use load_model_knn instead):

[classifier, MEAN, STD, classNames, mt_win, mt_step, st_win, st_step] =

aT.load_model(model_name)

# mid-term feature extraction:

[mt_features, _, _] = aF.mtFeatureExtraction(x, Fs, mt_win * Fs,

mt_step * Fs, round(Fs*st_win), round(Fs*st_step));

# feature normalization:

curFV = (mt_features[:, i] - MEAN) / STD;

# classification

[Result, P] = classifierWrapper(classifier, model_type, curFV)

"""

class_id = -1

probability = -1

if classifier_type == "knn":

class_id, probability = classifier.classify(test_sample)

elif classifier_type == "svm" or \

classifier_type == "randomforest" or \

classifier_type == "gradientboosting" or \

classifier_type == "extratrees" or \

classifier_type == "svm_rbf":

class_id = classifier.predict(test_sample.reshape(1, -1))[0]

probability = classifier.predict_proba(test_sample.reshape(1, -1))[0]

return class_id, probability