class Kernel(object):

@statistics
def linear():
    pass

@staticmethod
def gaussian(sigma):
    pass

@staticmethod
def polykernel(dimension, offset=0.0):
    pass

@staticmethod
def laplace(sigma):
    pass

@staticmethod
def hyperbolic_tangent(kappa, c):
    pass

class BaseSupportVector:

    @staticmethod
    def _get_kernel_func(kernel_name, kwargs):
        if kernel_name == KernelType.linear:
            return Kernel.linear()
        elif kernel_name == KernelType.gaussian:
            return Kernel.gaussian(kwargs['sigma'])
        elif kernel_name == KernelType.polynomial:
            if 'o' in kwargs:
                return Kernel.polykernel(kwargs['d'], kwargs['o'])
            else:
                return Kernel.polykernel(kwargs['d'])
        elif kernel_name == KernelType.laplace:
            return Kernel.laplace(kwargs['sigma'])
        elif kernel_name == KernelType.sigmoid:
            return Kernel.hyperbolic_tangent(kwargs['beta'], kwargs['theta'])
        else:
            raise KernelTypeError("非法的核函数名称")

class SVR(BaseSupportVector, BaseClassifier):

    __doc__ = "Support Vector Regression"

    def __init__(self, c, eps=0.1, kernel=KernelType.linear, **kwargs):
        """
        param:
            C           软间隔支持向量机参数（越大越迫使所有样本满足约束）
            eps         容忍的带宽
            kernel_type 核函数类型:
                        linear（无需提供参数，相当于没有用核函数）
                        polynomial(需提供参数：d)
                        gassian(需提供参数：sigma)
                        laplace(需提供参数：sigma)
                        sigmoid(需提供参数：beta, theta)
        """
        super(SVR, self).__init__()

def fit(self, x, y)

def predict(self, x)

def score(self, x, y)

def regression_plot(self, x):
    y = self.predict(x)
    regression_plot(self.x, self.y, x, y, title="SVR")

class SVM(BaseClassifier, BaseSupportVector, Multi2Binary):

    __doc__ = "Support Vector Machine"

    def __init__(self, c, eps=0.1, kernel=KernelType.linear, **kwargs):
        """
        param:
            C           软间隔支持向量机参数（越大越迫使所有样本满足约束）
            eps         容忍的带宽
            kernel_type 核函数类型:
                        linear（无需提供参数，相当于没有用核函数）
                        polynomial(需提供参数：d)
                        gassian(需提供参数：sigma)
                        laplace(需提供参数：sigma)
                        sigmoid(需提供参数：beta, theta)
        """
        super(SVM, self).__init__()

def fit(self, x, y)

def predict(self, x)

def score(self, x, y)

def classify_plot(self, x, y, title=""):
        classify_plot(self.new(), self.x, self.y, x, y, title=self.__doc__ + title)