使用 scikit-learn SVM 和 optunity 时出现“错误的输入形状”

Question

我正在尝试使用optunity package 来调整我的 SVM 模型，我直接复制并过去了它的最新示例代码，只需导入特征数组和数据数组

import optunity
import optunity.metrics
import sklearn.svm
import numpy as np

data_path = '/python/Feature'
files = ['A.npy', 'B.npy', 'C.npy']

array = []
labels = []

for i,name in enumerate(files):
    data = np.load('{}/{}'.format(data_path, name))
    for j in range(0,len(data)):
        labels.append(data[j])
        array.append(data)

print len(array)   #=> 1247
print len(labels)  #=> 1247

# score function: twice iterated 10-fold cross-validated accuracy
@optunity.cross_validated(x=data, y=labels, num_folds=10, num_iter=2)
def svm_auc(x_train, y_train, x_test, y_test, C, gamma):
    model = sklearn.svm.SVC(C=C, gamma=gamma).fit(x_train, y_train)
    decision_values = model.decision_function(x_test)
    return optunity.metrics.roc_auc(y_test, decision_values)

# perform tuning
optimal_pars, _, _ = optunity.maximize(svm_auc, num_evals=200, C=[0, 10], gamma=[0, 1])

# train model on the full training set with tuned hyperparameters
optimal_model = sklearn.svm.SVC(**optimal_pars).fit(data, labels)

但是，编译器看起来很不高兴，我查看了SVM class document 以仔细检查输入格式，但是我不明白 optunity 的编码语法.. 谁能帮我找出那里出了什么问题？非常感谢..（我正在使用'rbf'内核，我尝试添加但语法错误，在 optunity 的示例中很奇怪没有内核选择..）

Traceback (most recent call last):
  File "python/SVM_turning.py", line 26, in <module>
    optimal_pars, _, _ = optunity.maximize(svm_auc, num_evals=200, C=[0, 10], gamma=[0, 1])
  File "/lib/python2.7/site-packages/optunity/api.py", line 181, in maximize
    pmap=pmap)
  File "/lib/python2.7/site-packages/optunity/api.py", line 245, in optimize
    solution, report = solver.optimize(f, maximize, pmap=pmap)
  File "/lib/python2.7/site-packages/optunity/solvers/ParticleSwarm.py", line 257, in optimize
    fitnesses = pmap(evaluate, list(map(self.particle2dict, pop)))
  File "/lib/python2.7/site-packages/optunity/solvers/ParticleSwarm.py", line 246, in evaluate
    return f(**d)
  File "/lib/python2.7/site-packages/optunity/functions.py", line 286, in wrapped_f
    value = f(*args, **kwargs)
  File "/lib/python2.7/site-packages/optunity/functions.py", line 341, in wrapped_f
    return f(*args, **kwargs)
  File "/lib/python2.7/site-packages/optunity/constraints.py", line 150, in wrapped_f
    return f(*args, **kwargs)
  File "/lib/python2.7/site-packages/optunity/constraints.py", line 128, in wrapped_f
    return f(*args, **kwargs)
  File "/lib/python2.7/site-packages/optunity/constraints.py", line 265, in func
    return f(*args, **kwargs)
  File "/lib/python2.7/site-packages/optunity/cross_validation.py", line 386, in __call__
    scores.append(self.f(**kwargs))
  File "/python/SVM_turning.py", line 21, in svm_auc
    model = sklearn.svm.SVC(C=C, gamma=gamma).fit(x_train, y_train)
  File "/lib/python2.7/site-packages/sklearn/svm/base.py", line 138, in fit
    y = self._validate_targets(y)
  File "/lib/python2.7/site-packages/sklearn/svm/base.py", line 441, in _validate_targets
    y_ = column_or_1d(y, warn=True)
  File "/lib/python2.7/site-packages/sklearn/utils/validation.py", line 319, in column_or_1d
    raise ValueError("bad input shape {0}".format(shape))
ValueError: bad input shape (428, 600)

Answer 1

我想我找到了问题所在。您在读取文件时正在准备列表array 和labels。 array 依次被 data 填充。但是，稍后，您会这样做：

@optunity.cross_validated(x=data, y=labels, num_folds=10, num_iter=2)

和

optimal_model = sklearn.svm.SVC(**optimal_pars).fit(data, labels)

因此使用data 作为您的数据集，而不是您准备的array。我不知道你从文件中读取的格式，所以我不能确定发生了什么。但是，data 和 labels 的尺寸几乎肯定不会匹配。

这是一个使用 array 和 labels 的玩具示例，它们可以正常工作：

import optunity
import optunity.metrics
import sklearn.svm
import numpy as np

#print len(array)   #=> 1247
#print len(labels)  #=> 1247

# make dummy data
array = np.array([[i] for i in range(1247)])
labels = [True] * 100 + [False] * 1147

# score function: twice iterated 10-fold cross-validated accuracy
@optunity.cross_validated(x=array, y=labels, num_folds=10, num_iter=2)
def svm_auc(x_train, y_train, x_test, y_test, C, gamma):
    model = sklearn.svm.SVC(C=C, gamma=gamma).fit(x_train, y_train)
    decision_values = model.decision_function(x_test)
    return optunity.metrics.roc_auc(y_test, decision_values)

# perform tuning
optimal_pars, _, _ = optunity.maximize(svm_auc, num_evals=200, C=[0, 10], gamma=[0, 1])

# train model on the full training set with tuned hyperparameters
optimal_model = sklearn.svm.SVC(**optimal_pars).fit(array, labels)
print(optimal_pars)

哪些输出（示例）：

{'C': 8.0126953125, 'gamma': 0.35791015625}

抱歉这么久才回复。

Answer 2

我没有看到 optunity 中默认的优化器是什么，但是如果你只是使用网格搜索，你可以在 scikit-learn 中使用 GridSearchCV。

您的示例看起来很像 optunity 中的文档。您是否尝试过运行那里的确切示例？

Answer 3

不确定这是否是你的错误，但我在使用常规数组时得到了这个，而我应该使用 Numpy。