svm_SVC_使用GridSearchCV_癌症数据

超、凢脫俗 2022-05-20 01:24 178阅读 0赞

#scikit-learn中datasets自带的医学癌症数据
    #使用默认的高斯函数，使用GridSearchCV来自动选择参数gamma，可以得到在超参数调优器的使用下，得到的最优模型评分

from sklearn import  datasets
    datas=datasets.load_breast_cancer()
    print(datas)

运行结果：

{'DESCR': 'Breast Cancer Wisconsin (Diagnostic) Database\n=============================================\n\nNotes\n-----\nData Set Characteristics:\n    :Number of Instances: 569\n\n    :Number of Attributes: 30 numeric, predictive attributes and the class\n\n    :Attribute Information:\n        - radius (mean of distances from center to points on the perimeter)\n        - texture (standard deviation of gray-scale values)\n        - perimeter\n        - area\n        - smoothness (local variation in radius lengths)\n        - compactness (perimeter^2 / area - 1.0)\n        - concavity (severity of concave portions of the contour)\n        - concave points (number of concave portions of the contour)\n        - symmetry \n        - fractal dimension ("coastline approximation" - 1)\n\n        The mean, standard error, and "worst" or largest (mean of the three\n        largest values) of these features were computed for each image,\n        resulting in 30 features.  For instance, field 3 is Mean Radius, field\n        13 is Radius SE, field 23 is Worst Radius.\n\n        - class:\n                - WDBC-Malignant\n                - WDBC-Benign\n\n    :Summary Statistics:\n\n    ===================================== ====== ======\n                                           Min    Max\n    ===================================== ====== ======\n    radius (mean):                        6.981  28.11\n    texture (mean):                       9.71   39.28\n    perimeter (mean):                     43.79  188.5\n    area (mean):                          143.5  2501.0\n    smoothness (mean):                    0.053  0.163\n    compactness (mean):                   0.019  0.345\n    concavity (mean):                     0.0    0.427\n    concave points (mean):                0.0    0.201\n    symmetry (mean):                      0.106  0.304\n    fractal dimension (mean):             0.05   0.097\n    radius (standard error):              0.112  2.873\n    texture (standard error):             0.36   4.885\n    perimeter (standard error):           0.757  21.98\n    area (standard error):                6.802  542.2\n    smoothness (standard error):          0.002  0.031\n    compactness (standard error):         0.002  0.135\n    concavity (standard error):           0.0    0.396\n    concave points (standard error):      0.0    0.053\n    symmetry (standard error):            0.008  0.079\n    fractal dimension (standard error):   0.001  0.03\n    radius (worst):                       7.93   36.04\n    texture (worst):                      12.02  49.54\n    perimeter (worst):                    50.41  251.2\n    area (worst):                         185.2  4254.0\n    smoothness (worst):                   0.071  0.223\n    compactness (worst):                  0.027  1.058\n    concavity (worst):                    0.0    1.252\n    concave points (worst):               0.0    0.291\n    symmetry (worst):                     0.156  0.664\n    fractal dimension (worst):            0.055  0.208\n    ===================================== ====== ======\n\n    :Missing Attribute Values: None\n\n    :Class Distribution: 212 - Malignant, 357 - Benign\n\n    :Creator:  Dr. William H. Wolberg, W. Nick Street, Olvi L. Mangasarian\n\n    :Donor: Nick Street\n\n    :Date: November, 1995\n\nThis is a copy of UCI ML Breast Cancer Wisconsin (Diagnostic) datasets.\nhttps://goo.gl/U2Uwz2\n\nFeatures are computed from a digitized image of a fine needle\naspirate (FNA) of a breast mass.  They describe\ncharacteristics of the cell nuclei present in the image.\n\nSeparating plane described above was obtained using\nMultisurface Method-Tree (MSM-T) [K. P. Bennett, "Decision Tree\nConstruction Via Linear Programming." Proceedings of the 4th\nMidwest Artificial Intelligence and Cognitive Science Society,\npp. 97-101, 1992], a classification method which uses linear\nprogramming to construct a decision tree.  Relevant features\nwere selected using an exhaustive search in the space of 1-4\nfeatures and 1-3 separating planes.\n\nThe actual linear program used to obtain the separating plane\nin the 3-dimensional space is that described in:\n[K. P. Bennett and O. L. Mangasarian: "Robust Linear\nProgramming Discrimination of Two Linearly Inseparable Sets",\nOptimization Methods and Software 1, 1992, 23-34].\n\nThis database is also available through the UW CS ftp server:\n\nftp ftp.cs.wisc.edu\ncd math-prog/cpo-dataset/machine-learn/WDBC/\n\nReferences\n----------\n   - W.N. Street, W.H. Wolberg and O.L. Mangasarian. Nuclear feature extraction \n     for breast tumor diagnosis. IS&T/SPIE 1993 International Symposium on \n     Electronic Imaging: Science and Technology, volume 1905, pages 861-870,\n     San Jose, CA, 1993.\n   - O.L. Mangasarian, W.N. Street and W.H. Wolberg. Breast cancer diagnosis and \n     prognosis via linear programming. Operations Research, 43(4), pages 570-577, \n     July-August 1995.\n   - W.H. Wolberg, W.N. Street, and O.L. Mangasarian. Machine learning techniques\n     to diagnose breast cancer from fine-needle aspirates. Cancer Letters 77 (1994) \n     163-171.\n', 'target_names': array(['malignant', 'benign'],
          dtype='<U9'), 'data': array([[  1.79900000e+01,   1.03800000e+01,   1.22800000e+02, ...,
              2.65400000e-01,   4.60100000e-01,   1.18900000e-01],
           [  2.05700000e+01,   1.77700000e+01,   1.32900000e+02, ...,
              1.86000000e-01,   2.75000000e-01,   8.90200000e-02],
           [  1.96900000e+01,   2.12500000e+01,   1.30000000e+02, ...,
              2.43000000e-01,   3.61300000e-01,   8.75800000e-02],
           ..., 
           [  1.66000000e+01,   2.80800000e+01,   1.08300000e+02, ...,
              1.41800000e-01,   2.21800000e-01,   7.82000000e-02],
           [  2.06000000e+01,   2.93300000e+01,   1.40100000e+02, ...,
              2.65000000e-01,   4.08700000e-01,   1.24000000e-01],
           [  7.76000000e+00,   2.45400000e+01,   4.79200000e+01, ...,
              0.00000000e+00,   2.87100000e-01,   7.03900000e-02]]), 'feature_names': array(['mean radius', 'mean texture', 'mean perimeter', 'mean area',
           'mean smoothness', 'mean compactness', 'mean concavity',
           'mean concave points', 'mean symmetry', 'mean fractal dimension',
           'radius error', 'texture error', 'perimeter error', 'area error',
           'smoothness error', 'compactness error', 'concavity error',
           'concave points error', 'symmetry error', 'fractal dimension error',
           'worst radius', 'worst texture', 'worst perimeter', 'worst area',
           'worst smoothness', 'worst compactness', 'worst concavity',
           'worst concave points', 'worst symmetry', 'worst fractal dimension'],
          dtype='<U23'), 'target': array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 0,
           0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0,
           1, 0, 1, 1, 1, 1, 1, 0, 0, 1, 0, 0, 1, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1,
           1, 0, 1, 0, 0, 1, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 0, 0, 1, 1, 1, 0,
           1, 1, 0, 0, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1,
           1, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 1, 1,
           0, 1, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1,
           0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 1, 1, 0, 1, 1, 0, 0, 0, 1,
           0, 1, 0, 1, 1, 1, 0, 1, 1, 0, 0, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 1,
           0, 1, 1, 0, 1, 0, 0, 0, 0, 1, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0,
           0, 1, 1, 0, 1, 1, 0, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 0,
           0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1,
           1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1,
           1, 0, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0,
           1, 0, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 0, 1, 0, 1, 0, 1, 1, 1, 0, 1,
           1, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1,
           0, 0, 0, 1, 0, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1,
           1, 0, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1,
           0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 0, 1, 0, 1,
           1, 1, 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1,
           0, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1,
           1, 1, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 0, 1, 1,
           1, 1, 1, 0, 1, 1, 0, 1, 0, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1,
           1, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
           1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 1])}

from sklearn.model_selection import train_test_split,GridSearchCV
    from sklearn import metrics
    from sklearn.svm import SVC
    import numpy as np

x_train,x_test,y_train,y_test=train_test_split(datas.data,datas.target)

thresholds=np.linspace(0,0.001,100)#设置gamma参数列表
    thresholds

运行结果：

array([  0.00000000e+00,   1.01010101e-05,   2.02020202e-05,
             3.03030303e-05,   4.04040404e-05,   5.05050505e-05,
             6.06060606e-05,   7.07070707e-05,   8.08080808e-05,
             9.09090909e-05,   1.01010101e-04,   1.11111111e-04,
             1.21212121e-04,   1.31313131e-04,   1.41414141e-04,
             1.51515152e-04,   1.61616162e-04,   1.71717172e-04,
             1.81818182e-04,   1.91919192e-04,   2.02020202e-04,
             2.12121212e-04,   2.22222222e-04,   2.32323232e-04,
             2.42424242e-04,   2.52525253e-04,   2.62626263e-04,
             2.72727273e-04,   2.82828283e-04,   2.92929293e-04,
             3.03030303e-04,   3.13131313e-04,   3.23232323e-04,
             3.33333333e-04,   3.43434343e-04,   3.53535354e-04,
             3.63636364e-04,   3.73737374e-04,   3.83838384e-04,
             3.93939394e-04,   4.04040404e-04,   4.14141414e-04,
             4.24242424e-04,   4.34343434e-04,   4.44444444e-04,
             4.54545455e-04,   4.64646465e-04,   4.74747475e-04,
             4.84848485e-04,   4.94949495e-04,   5.05050505e-04,
             5.15151515e-04,   5.25252525e-04,   5.35353535e-04,
             5.45454545e-04,   5.55555556e-04,   5.65656566e-04,
             5.75757576e-04,   5.85858586e-04,   5.95959596e-04,
             6.06060606e-04,   6.16161616e-04,   6.26262626e-04,
             6.36363636e-04,   6.46464646e-04,   6.56565657e-04,
             6.66666667e-04,   6.76767677e-04,   6.86868687e-04,
             6.96969697e-04,   7.07070707e-04,   7.17171717e-04,
             7.27272727e-04,   7.37373737e-04,   7.47474747e-04,
             7.57575758e-04,   7.67676768e-04,   7.77777778e-04,
             7.87878788e-04,   7.97979798e-04,   8.08080808e-04,
             8.18181818e-04,   8.28282828e-04,   8.38383838e-04,
             8.48484848e-04,   8.58585859e-04,   8.68686869e-04,
             8.78787879e-04,   8.88888889e-04,   8.98989899e-04,
             9.09090909e-04,   9.19191919e-04,   9.29292929e-04,
             9.39393939e-04,   9.49494949e-04,   9.59595960e-04,
             9.69696970e-04,   9.79797980e-04,   9.89898990e-04,
             1.00000000e-03])

#这步，直接影响下面的数据
    param_grid={'gamma':thresholds}
    clf=GridSearchCV(SVC(kernel='rbf'),param_grid,cv=5)
    clf.fit(x_train,y_train)

运行结果：

GridSearchCV(cv=5, error_score='raise',
           estimator=SVC(C=1.0, cache_size=200, class_weight=None, coef0=0.0,
      decision_function_shape='ovr', degree=3, gamma='auto', kernel='rbf',
      max_iter=-1, probability=False, random_state=None, shrinking=True,
      tol=0.001, verbose=False),
           fit_params=None, iid=True, n_jobs=1,
           param_grid={'gamma': array([  0.00000e+00,   1.01010e-05, ...,   9.89899e-04,   1.00000e-03])},
           pre_dispatch='2*n_jobs', refit=True, return_train_score=True,
           scoring=None, verbose=0)

print("最佳效果：%0.3f"% clf.best_score_)
    print("最优参数组合：")
    best_parameters=clf.best_estimator_.get_params()
    for param_name in sorted(param_grid.keys()):
        print('\t%s:%r' %(param_name,best_parameters[param_name]))

运行结果：

最佳效果：0.925
    最优参数组合：
    	gamma:8.0808080808080811e-05

注：这步结果不唯一，下面的数据也不一致，原因详见https://blog.csdn.net/wjwfighting/article/details/80970396，开头有提到

print("训练集评分：",clf.score(x_train,y_train))
    print("测试集评分：",clf.score(x_test,y_test))

训练集评分： 0.934272300469
    测试集评分： 0.979020979021

predicted=clf.predict(x_test)
    print('预测值：',predicted)
    print('实际值:',y_test)

预测值： [0 1 0 1 1 1 1 0 0 1 0 0 0 0 1 1 0 0 1 0 1 1 1 1 0 1 0 1 0 0 1 1 1 1 0 0 1
     1 1 1 1 1 1 0 1 1 0 1 1 1 0 1 0 1 1 1 1 1 1 1 0 1 1 1 0 1 1 0 1 1 1 0 0 1
     1 0 1 0 0 1 1 0 0 1 0 0 0 0 0 1 1 1 0 1 0 1 1 1 1 1 0 1 1 1 1 0 1 1 1 1 1
     1 1 0 1 1 1 0 1 0 1 0 1 1 0 0 1 1 1 1 0 1 1 0 1 1 1 1 1 0 0 0 0]
    实际值: [0 1 0 1 1 1 1 0 0 1 0 0 0 0 1 1 0 0 1 0 1 1 1 1 0 1 0 1 0 0 1 1 1 1 0 0 1
     1 1 1 1 1 1 0 1 1 0 1 1 1 0 1 0 1 1 0 1 1 1 1 0 1 1 1 1 1 1 0 1 1 1 0 0 1
     1 0 1 0 0 1 1 0 0 1 0 0 0 0 0 1 1 0 0 1 0 1 1 1 1 1 0 1 1 1 1 0 1 1 1 1 1
     1 1 0 1 1 1 0 1 0 1 0 1 1 0 0 1 1 1 1 0 1 1 0 1 1 1 1 1 0 0 0 0]

print('精准值：',metrics.precision_score(predicted,y_test))
    print('召回率：',metrics.recall_score(predicted,y_test))
    print('F1:',metrics.f1_score(predicted,y_test))
    print("准确率:",np.mean(predicted==y_test))

运行结果：

精准值： 0.988888888889
    召回率： 0.978021978022
    F1: 0.983425414365
    准确率: 0.979020979021

#分类报告
    xx=metrics.classification_report(y_test,predicted,target_names=datas.target_names)
    print(xx)

运行结果：

precision    recall  f1-score   support
    
      malignant       0.98      0.96      0.97        53
         benign       0.98      0.99      0.98        90
    
    avg / total       0.98      0.98      0.98       143

confusion_matrix=metrics.confusion_matrix(y_test,predicted)
    print('--混淆矩阵--')
    print(confusion_matrix)

运行结果：

--混淆矩阵--
    [[51  2]
     [ 1 89]]