Python 第三方模块 机器学习 Scikit-Learn模块 有监督学习1 交叉分解,高斯过程,保序回归

一.cross_decomposition
1.简介:

该模块用于进行"交叉分解"(cross decomposition)

2.使用:

"典型相关分析"(Canonical Correlation Analysis;CCA):class sklearn.cross_decomposition.CCA([n_components=2,scale=True,max_iter=500,tol=1e-06,copy=True])
  #参数说明:
	n_components:指定要保留的组件数;为int
	scale:指定是否缩放数据;为bool
	max_iter:指定NIPALS内部循环的最大迭代次数;为int
	tol:指定最小误差(若误差小于该值,则停止);为float>=0
	copy:指定是否复制数据;为bool

######################################################################################################################

"偏最小二乘"(Partial Least Squares;PLS)转换与回归:class sklearn.cross_decomposition.PLSCanonical([n_components=2,scale=True,algorithm='nipals',max_iter=500,tol=1e-06,copy=True])
  #参数说明:其他参数同class sklearn.cross_decomposition.CCA()
	algorithm:指定用于估计"互协方差矩阵"(cross-covariance matrix)的第1个奇异向量的算法;为"nipals"/"svd"

######################################################################################################################

偏最小二乘回归:class sklearn.cross_decomposition.PLSRegression([n_components=2,scale=True,max_iter=500,tol=1e-06,copy=True])
  #参数说明:同class sklearn.cross_decomposition.CCA()

######################################################################################################################

偏最小二乘"奇异值分解"(Singular Value Decomposition;SVD):class sklearn.cross_decomposition.PLSSVD([n_components=2,scale=True,copy=True])
  #参数说明:同class sklearn.cross_decomposition.CCA()

二.gaussian_process
1.简介:

该模块实现了基于"高斯过程"(Gaussian Process;GP)的回归和分类

2.使用:

基于"拉普拉斯近似"(Laplace approximation)的"高斯过程分类"(Gaussian process classification;GPC):class sklearn.gaussian_process.GaussianProcessClassifier([kernel=None,optimizer='fmin_l_bfgs_b',n_restarts_optimizer=0,max_iter_predict=100,warm_start=False,copy_X_train=True,random_state=None,multi_class='one_vs_rest',n_jobs=None])
  #参数说明:
	kernel:指定GP的协方差函数的核;为kernel instance
	optimizer:指定优化器,用于优化内核参数;为"fmin_l_bfgs_b"/callable
	n_restarts_optimizer:指定用于查找[使"对数边际似然"(log-marginal likelihood)最大化的内核参数]的优化器重启的次数;为int
	max_iter_predict:指定用于近似"后验概率"(Posterior)的牛顿法的最大迭代次数;为int
	warm_start:指定是否启用"热重启"(warm start);为bool
	copy_X_train:指定是否复制数据;为bool
	random_state:指定用于初始化中心的随机数;为int/RandomState instance/None
	multi_class:指定如何处理多类分类问题;为"one_vs_rest"/"one_vs_one"
	n_jobs:指定用于计算的作业数量;为int

######################################################################################################################

"高斯过程回归"(Gaussian process regression;GPR):class sklearn.gaussian_process.GaussianProcessRegressor([kernel=None,alpha=1e-10,optimizer='fmin_l_bfgs_b',n_restarts_optimizer=0,normalize_y=False,copy_X_train=True,random_state=None])
  #参数说明:其他参数同class sklearn.gaussian_process.GaussianProcessClassifier()
	alpha:指定拟合时在核矩阵对角线上增加的值;为float/1×n_samples array-like
	normalize_y:指定是否对目标值y进行归一化;为bool

3.gaussian_process.kernels
(1)简介:

该子模块定义了高斯过程的"核函数"(kernel function)

(2)使用:

所有核的基类:class sklearn.gaussian_process.kernels.Kernel

######################################################################################################################

由1组其他核组成的核:class sklearn.gaussian_process.kernels.CompoundKernel(<kernels>)

######################################################################################################################

"常数核"(Constant kernel):class sklearn.gaussian_process.kernels.ConstantKernel([constant_value=1.0,constant_value_bounds=(1e-05,100000.0)])
"点积核"(Dot-Product kernel):class sklearn.gaussian_process.kernels.DotProduct([sigma_0=1.0,sigma_0_bounds=(1e-05,100000.0)])
"指数正弦平方核"(Exp-Sine-Squared kernel)/"周期核"(periodic kernel):class sklearn.gaussian_process.kernels.ExpSineSquared([length_scale=1.0,periodicity=1.0,length_scale_bounds=(1e-05,100000.0),periodicity_bounds=(1e-05,100000.0)])
"指数核"(Exponentiation kernel):class sklearn.gaussian_process.kernels.Exponentiation(<kernel>,<exponent>)
"马顿核"(Matern kernel):class sklearn.gaussian_process.kernels.Matern([length_scale=1.0,length_scale_bounds=(1e-05,100000.0),nu=1.5])
对sklearn.metrics.pairwise中的内核的封装:class sklearn.gaussian_process.kernels.PairwiseKernel([gamma=1.0,gamma_bounds=(1e-05,100000.0),metric='linear',pairwise_kernels_kwargs=None])
"乘积核"(Product kernel):class sklearn.gaussian_process.kernels.Product(<k1>,<k2>)
"径向基函数核"(Radial-basis function kernel;RBF kernel)/"平方指数核"(squared-exponential kernel;SE kernel):class sklearn.gaussian_process.kernels.RBF([length_scale=1.0,length_scale_bounds=(1e-05,100000.0)])
"有理二次核"(Rational Quadratic kernel;RQ kernel):class sklearn.gaussian_process.kernels.RationalQuadratic([length_scale=1.0,alpha=1.0,length_scale_bounds=(1e-05,100000.0),alpha_bounds=(1e-05,100000.0)])
"求和核"(Sum kernel):class sklearn.gaussian_process.kernels.Sum(<k1>,<k2>)
"白核"(White kernel):class sklearn.gaussian_process.kernels.WhiteKernel([noise_level=1.0,noise_level_bounds=(1e-05,100000.0)])

######################################################################################################################

A kernel hyperparameter's specification in form of a namedtuple:class sklearn.gaussian_process.kernels.Hyperparameter(<name>,<value_type>,<bounds>[,n_elements=1,fixed=None])

三.isotonic
1.简介:

该模块用于进行"保序回归"(Isotonic regression)

2.使用
(1)类:

"保序回归模型"(Isotonic regression model):class sklearn.isotonic.IsotonicRegression([y_min=None,y_max=None,increasing=True,out_of_bounds='nan'])

(2)函数:

求因变量是否与自变量单调相关:[<increasing_bool>=]sklearn.isotonic.check_increasing(<x>,<y>)
  #参数说明:
	x,y:分别指定自/因变量;均为1×n_samples array-like
	increasing_bool:返回结果;为bool

######################################################################################################################

求解保序回归模型:[<y_>=]sklearn.isotonic.isotonic_regression(<y>[,sample_weight=None,y_min=None,y_max=None,increasing=True])