Sklearn(v3)——SVM理论(2)

最新推荐文章于 2024-04-02 06:39:40 发布

Grateful_Dead424

最新推荐文章于 2024-04-02 06:39:40 发布

阅读量723

点赞数

文章标签： sklearn 机器学习 SVM

本文链接：https://blog.csdn.net/Grateful_Dead424/article/details/121506840

版权

from sklearn.datasets import make_blobs
from sklearn.svm import SVC
import matplotlib.pyplot as plt
import numpy as np

X,y = make_blobs(n_samples=50,centers=2,random_state=0,cluster_std=0.6)

plt.scatter(X[:,0],X[:,1],c=y,s=50,cmap="rainbow")
plt.xticks([])
plt.yticks([])
plt.show()

#获取平面上两条坐标轴的最大值和最小值
xlim = ax.get_xlim()
ylim = ax.get_ylim()
#在最大值和最小值之间形成30个规律的数据
axisx = np.linspace(xlim[0],xlim[1],30)
axisy = np.linspace(ylim[0],ylim[1],30)
axisx

array([-0.7425579 , -0.60066997, -0.45878204, -0.31689411, -0.17500618,
       -0.03311826,  0.10876967,  0.2506576 ,  0.39254553,  0.53443346,
        0.67632139,  0.81820931,  0.96009724,  1.10198517,  1.2438731 ,
        1.38576103,  1.52764896,  1.66953689,  1.81142481,  1.95331274,
        2.09520067,  2.2370886 ,  2.37897653,  2.52086446,  2.66275238,
        2.80464031,  2.94652824,  3.08841617,  3.2303041 ,  3.37219203])

axisy,axisx = np.meshgrid(axisy,axisx)
axisx #我们将使用这里形成的二维数组作为我们contour函数中的X和Y 
#使用meshgrid函数将两个一维向量转换为特征矩阵
#核心是将两个特征向量广播，以便获取y.shape * x.shape这么多个坐标点的横坐标和纵坐标

array([[-0.7425579 , -0.7425579 , -0.7425579 , ..., -0.7425579 ,
        -0.7425579 , -0.7425579 ],
       [-0.7425579 , -0.7425579 , -0.7425579 , ..., -0.7425579 ,
        -0.7425579 , -0.7425579 ],
       [-0.7425579 , -0.7425579 , -0.7425579 , ..., -0.7425579 ,
        -0.7425579 , -0.7425579 ],
       ...,
       [ 3.37219203,  3.37219203,  3.37219203, ...,  3.37219203,
         3.37219203,  3.37219203],
       [ 3.37219203,  3.37219203,  3.37219203, ...,  3.37219203,
         3.37219203,  3.37219203],
       [ 3.37219203,  3.37219203,  3.37219203, ...,  3.37219203,
         3.37219203,  3.37219203]])

xy = np.vstack([axisx.ravel(), axisy.ravel()]).T  #其中ravel()是降维函数，vstack能够将多个结构一致的一维数组按行堆叠起来
#xy就是已经形成的网格，它是遍布在整个画布上的密集的点
plt.scatter(xy[:,0],xy[:,1],s=1,cmap="rainbow") #理解函数meshgrid和vstack的作用

plt.scatter(X[:,0],X[:,1],c=y,s=50,cmap="rainbow")
ax = plt.gca()
#建模，通过fit计算出对应的决策边界
clf = SVC(kernel="linear").fit(X,y)
Z = clf.decision_function(xy).reshape(axisx.shape)
#重要接口decision_function，返回每个输入的样本所对应的到决策边界的距离
#然后再将这个距离转换为axisx的结构，这是由于画图的函数contour要求Z的结构必须与X和Y保持一致
#画决策边界和平行于决策边界的超平面
ax.contour(axisx,axisy,Z,colors="k",levels=[-1,0,1],alpha=0.5,linestyles=["--","-","--"])  #画三条等高线，分别是Z为-1，Z为0和Z为1的三条线

ax.set_xlim(xlim)
ax.set_ylim(ylim)

#记得Z的本质么？是输入的样本到决策边界的距离，而contour函数中的level其实是输入了这个距离
#让我们用一个点来试试看
plt.scatter(X[:,0],X[:,1],c=y,s=50,cmap="rainbow")
plt.scatter(X[10,0],X[10,1],c="black",s=50,cmap="rainbow")

clf.decision_function(X[10].reshape(1,2))

结果：

array([-3.33917354])

plt.scatter(X[:,0],X[:,1],c=y,s=50,cmap="rainbow")
ax = plt.gca()
ax.contour(axisx,axisy,Z,colors="k",levels=[-3.33917354],alpha=0.5,linestyles=["--"])

#将上述过程包装成函数
def plot_svc_decision_function(model,ax=None):
    if ax is None:
        ax = plt.gca()
    xlim = ax.get_xlim()
    ylim = ax.get_ylim()
    x = np.linspace(xlim[0],xlim[1],30)
    y = np.linspace(ylim[0],ylim[1],30)
    Y,X = np.meshgrid(y,x) 
    xy = np.vstack([X.ravel(), Y.ravel()]).T
    P = model.decision_function(xy).reshape(X.shape)
    ax.contour(X, Y, P,colors="k",levels=[-1,0,1],alpha=0.5,linestyles=["--","-","--"]) 
    ax.set_xlim(xlim)
    ax.set_ylim(ylim)

clf = SVC(kernel = "linear").fit(X,y)    
plt.scatter(X[:,0],X[:,1],c=y,s=50,cmap="rainbow")
plot_svc_decision_function(clf)

clf.predict(X)

clf.score(X,y)

print(clf.support_vectors_)
#返回支持向量
clf.n_support_
#返回每个类中支持向量的个数

结果：

array([1, 1, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 1,
       1, 1, 1, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 1, 1, 0, 1, 0, 1, 0, 1, 1,
       0, 1, 1, 0, 1, 0])

1.0

[[0.44359863 3.11530945]
 [2.33812285 3.43116792]
 [2.06156753 1.96918596]]

array([2, 1], dtype=int32)

使用环形数据

from sklearn.datasets import make_circles
X,y = make_circles(100,factor=0.1,noise=.1)
plt.scatter(X[:,0],X[:,1],c=y,s=50,cmap="rainbow")
plt.show()

from sklearn.datasets import make_circles
X,y = make_circles(100,factor=0.1,noise=.1)
clf = SVC(kernel = "linear").fit(X,y)    
plt.scatter(X[:,0],X[:,1],c=y,s=50,cmap="rainbow")
plot_svc_decision_function(clf)

很明显，现在线性SVM已经不适合于我们的状况了，我们无法找出一条直线来划分我们的数据集，让直线的两边分别是两种类别。这个时候，如果我们能够在原本的X和y的基础上，添加一个维度r，变成三维，我们可视化这个数据，来看看添加维度让我们的数据如何变化。

r = np.exp(-(X**2).sum(1))
rlim = np.linspace(min(r),max(r),100)

from mpl_toolkits import mplot3d
def plot_3D(elev=30,azim=30,X=X,y=y):
    ax = plt.subplot(projection="3d")
    ax.scatter3D(X[:,0],X[:,1],r,c=y,s=50,cmap='rainbow')
    ax.view_init(elev=elev,azim=azim)
    ax.set_xlabel("x")
    ax.set_ylabel("y")
    ax.set_zlabel("r")
    plt.show()
from ipywidgets import interact,fixed
interact(plot_3D,elev=[0,30],azip=(-180,180),X=fixed(X),y=fixed(y))
plt.show()

此时我们的数据在三维空间中，我们的超平面就是一个二维平面。明显我们可以用一个平面将两类数据隔开，这个平面就是我们的决策边界了。我们刚才做的，计算r，并将r作为数据的第三维度来将数据升维的过程，被称为“核变换”，即是将数据投影到高维空间中，以寻找能够将数据完美分割的超平面，即是说寻找能够让数据线性可分的高维空间。为了详细解释这个过程，我们需要引入SVM中的核心概念：核函数。

clf = SVC(kernel = "rbf").fit(X,y)    
plt.scatter(X[:,0],X[:,1],c=y,s=50,cmap="rainbow")
plot_svc_decision_function(clf)

import numpy as np
import matplotlib.pyplot as plt
from matplotlib.colors import ListedColormap
from sklearn import svm
from sklearn.datasets import make_circles , make_moons, make_blobs,make_classification

n_samples = 100
datasets = [
    make_moons(n_samples=n_samples, noise=0.2, random_state=0),              
    make_circles(n_samples=n_samples, noise=0.2, factor=0.5, random_state=1), 
    make_blobs(n_samples=n_samples, centers=2, random_state=5),
    make_classification(n_samples=n_samples,n_features =2,n_informative=2,n_redundant=0, random_state=5)
]

Kernel = ["linear","poly","rbf","sigmoid"]

#四个数据集分别是什么样子呢？
for X,Y in datasets:
    plt.figure(figsize=(5,4))
    plt.scatter(X[:,0],X[:,1],c=Y,s=50,cmap="rainbow")

[*enumerate(datasets)]
#index,(X,Y) = [(索引，array([特征矩阵X]，[标签Y]))]

结果：

[(0,
  (array([[-1.09443462e-02,  9.89784876e-01],
          [ 1.96749886e+00, -1.10921978e-01],
          [ 9.18923151e-01, -7.87831621e-03],
          [-1.97813183e-02,  3.67422878e-02],
          [ 8.97047211e-01, -5.26043067e-01],
          [ 2.05087697e+00,  4.82966687e-01],
          [ 5.52592656e-01,  5.10008493e-01],
          [ 9.36108682e-01, -6.67176177e-01],
          [-8.57905150e-03,  3.44030710e-01],
          [ 1.79962867e+00,  3.22578165e-01],
          [-1.79739813e-01,  5.12417381e-01],
          [ 1.96928635e+00, -1.84060982e-01],
          [ 1.40757108e+00, -6.55885144e-01],
          [ 1.04103920e+00,  1.04537944e+00],
          [ 6.11861752e-01,  5.09315861e-01],
          [-3.59476500e-01,  1.05930036e+00],
          [ 2.54029695e-01,  1.15116524e+00],
          [ 2.13555501e-01,  8.82321641e-01],
          [-3.30880800e-01,  8.04221145e-01],
          [ 1.06603845e+00,  5.32174106e-01],
          [-7.06988363e-01,  5.63246401e-01],
          [ 4.02928450e-01, -1.95330382e-01],
          [ 7.83527128e-01,  5.65637444e-01],
          [ 7.29264348e-01, -4.64258931e-01],
          [-7.61131674e-01, -1.74321350e-03],
          [-8.30401440e-01, -2.33952062e-01],
          [ 7.12873757e-01,  3.33441281e-01],
          [ 2.12091446e+00,  1.51388354e-01],
          [ 1.76738365e+00, -1.38842428e-02],
          [ 1.35429861e+00, -2.35239859e-01],
          [ 3.82226943e-01, -1.29870625e-01],
          [ 1.15255238e+00, -8.36624186e-01],
          [ 1.85603425e+00, -2.25641253e-02],
          [ 4.78053620e-01, -3.54658215e-01],
          [ 4.65065876e-02,  5.22966374e-01],
          [-1.68749515e-01,  9.97161466e-01],
          [ 2.17677252e-01,  9.71890153e-01],
          [ 1.45168696e-01,  2.06362619e-01],
          [-6.04440255e-02,  4.86891449e-02],
          [ 1.00652060e+00, -5.83659180e-01],
          [ 1.34599608e+00, -8.74713518e-03],
          [ 5.07344926e-01, -3.11872588e-01],
          [-8.84426881e-01,  1.75672048e-01],
          [-1.00353955e+00,  2.54679349e-01],
          [ 1.00682339e+00,  3.36434579e-01],
          [ 8.11581056e-01,  1.19684303e+00],
          [ 6.05383054e-01,  1.34346598e+00],
          [-5.25267589e-01,  6.67755643e-01],
          [-9.36918623e-01,  3.24010896e-01],
          [ 8.32721148e-01,  2.07541427e-01],
          [ 1.56011397e+00, -1.61052076e-03],
          [-2.00343863e-01, -1.71769945e-01],
          [ 8.14368163e-01,  2.98144383e-01],
          [-5.33016793e-01,  7.25851388e-01],
          [ 1.39949996e-01,  5.16100416e-01],
          [ 1.30241869e-01,  2.73900710e-01],
          [ 6.05976627e-01,  8.71416086e-01],
          [-3.55599199e-01,  4.28344752e-01],
          [ 1.80905518e-01,  1.21324092e+00],
          [-6.86271500e-02,  4.98563121e-01],
          [ 6.91482442e-01,  7.02335678e-01],
          [-3.83113433e-01,  9.66746318e-01],
          [ 2.98656366e-01, -1.83495206e-01],
          [ 1.17897990e-01, -2.31064511e-01],
          [ 9.04410734e-01, -6.86183692e-01],
          [ 1.27108202e+00, -3.39556126e-01],
          [-2.52941845e-01,  9.36590815e-01],
          [ 1.58149755e+00, -5.26620862e-01],
          [ 7.04126938e-01,  6.45019632e-01],
          [ 2.05387806e+00, -4.99221849e-01],
          [ 2.78958975e-01,  8.79248341e-01],
          [-7.28199738e-01,  9.21967277e-01],
          [-9.21538389e-01,  4.83269613e-02],
          [ 2.01257720e+00,  2.06208601e-01],
          [ 2.09649727e+00,  4.53952338e-01],
          [ 4.55121438e-01, -5.98476065e-01],
          [ 3.25942701e-01,  1.06336046e+00],
          [ 1.80917678e+00,  3.67632943e-01],
          [ 7.05843148e-01, -4.60516884e-01],
          [ 1.24497910e+00, -4.89751662e-01],
          [-1.02501239e-01,  1.16337954e+00],
          [-6.13951804e-01,  9.35134524e-01],
          [ 1.32828610e+00, -2.76080239e-01],
          [-9.10782155e-01,  4.00675696e-01],
          [-8.86192869e-01, -1.04843093e-01],
          [ 1.59251994e-01, -3.50710293e-02],
          [ 1.74397698e-01, -4.22503039e-02],
          [ 7.28423493e-01,  4.19640376e-01],
          [ 1.70911154e+00, -1.84104334e-01],
          [ 5.76918298e-01, -2.31102480e-01],
          [ 1.68451790e+00, -3.48465285e-01],
          [-7.71915511e-01,  6.67699774e-01],
          [ 2.75312996e-01,  8.48999530e-01],
          [ 1.42394089e+00, -6.57814970e-01],
          [ 1.05961304e+00,  5.80782652e-01],
          [ 1.79348845e+00,  4.94760237e-01],
          [ 3.24902726e-01, -5.63880560e-01],
          [ 7.60779589e-01, -3.73787587e-01],
          [-1.20959690e+00,  3.24700620e-01],
          [-1.18089701e+00,  2.27800799e-01]]),
   array([0, 1, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1,
          0, 1, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0, 1, 0, 0,
          0, 0, 0, 0, 0, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 1, 1, 1,
          0, 1, 0, 1, 0, 0, 0, 1, 1, 1, 0, 1, 1, 1, 0, 0, 1, 0, 0, 1, 1, 0,
          1, 1, 1, 0, 0, 1, 0, 1, 1, 1, 0, 0]))),
 (1,
  (array([[-0.38289117, -0.09084004],
          [-0.02096229, -0.47787416],
          [-0.39611596, -1.28942694],
          [-0.61813018, -0.06383715],
          [ 0.70347838, -0.18703837],
          [-0.45970463,  0.69477465],
          [-0.45091682, -0.71570524],
          [-0.45562004, -0.13406016],
          [-0.26513904,  0.40812871],
          [-0.15474648,  0.41406973],
          [ 0.231206  , -0.53275899],
          [ 0.15623875, -0.8678088 ],
          [ 0.51647541,  0.48940995],
          [ 0.68707007, -0.02334129],
          [ 0.54759869, -0.16482373],
          [-0.32179572, -0.80553536],
          [-1.07478639, -0.733362  ],
          [ 0.76758455, -0.43498783],
          [-0.47475234, -0.33813186],
          [ 0.15420656,  1.07306032],
          [ 0.65916696,  0.20773634],
          [-0.77795003,  0.1326555 ],
          [ 0.47025454, -0.31228748],
          [-0.04246799,  0.26555446],
          [-0.72405954,  0.48807185],
          [-0.36960005, -1.06514028],
          [ 0.17833327, -0.49718972],
          [-0.93927864, -0.41951638],
          [ 0.50914152, -0.70977467],
          [-0.05569852, -0.82162607],
          [-0.11214579,  0.72197044],
          [ 0.80463921, -0.15221296],
          [ 0.08261487, -0.11749021],
          [ 0.20349541, -0.37396789],
          [ 0.13864693, -0.23905642],
          [ 0.32785307, -1.00769037],
          [ 0.88944061, -0.39117628],
          [-0.05837947,  0.28487039],
          [-1.0673653 ,  0.2204006 ],
          [-0.60071345, -0.69545189],
          [-0.03972324, -0.40936056],
          [ 0.39742085,  0.20621162],
          [-0.36941154,  0.0129811 ],
          [ 0.03573703,  0.46666229],
          [-0.56814999, -0.41288419],
          [ 0.41047299, -0.73640868],
          [ 0.88249707, -0.69004404],
          [ 0.06579822, -0.50458395],
          [-0.75737223, -0.0724028 ],
          [ 0.18316966,  0.08722007],
          [ 0.67248314, -0.41892665],
          [ 0.25898723,  0.39688645],
          [-1.1312983 ,  0.4810614 ],
          [ 1.0592844 ,  0.64490287],
          [ 0.41019663,  0.38790198],
          [ 0.95142029, -0.04089983],
          [-0.60492988,  0.43950906],
          [ 0.23314762, -0.81785711],
          [ 0.91067331,  0.30702075],
          [-0.45026472, -0.03724104],
          [-0.81396121, -0.64733959],
          [-0.23191338,  0.50533992],
          [-0.59760983,  0.28023168],
          [ 0.73960166, -0.84270281],
          [ 0.57294659, -0.31198928],
          [ 0.24821133, -0.54784509],
          [ 0.52127802,  0.94108005],
          [ 0.33973198,  0.10609978],
          [ 1.05339036, -0.02197593],
          [ 0.01327466, -0.63379502],
          [ 0.2422589 ,  0.49032064],
          [-0.89266612,  0.6345076 ],
          [ 0.1672566 ,  0.23548462],
          [-0.05611705,  0.38834099],
          [ 0.84695486,  0.81435811],
          [ 0.29976195, -0.07943031],
          [-0.1404762 ,  0.72486032],
          [-0.05482024,  0.18417328],
          [-0.24643884, -0.43283337],
          [-0.23460645,  0.6409442 ],
          [-1.13184893, -0.61964942],
          [-0.92413821, -0.45302089],
          [ 0.2225745 ,  0.77052597],
          [-0.69453765,  0.53014147],
          [-1.0362509 ,  0.77339965],
          [ 0.51880585,  0.30152232],
          [-0.77429541,  0.02553767],
          [ 0.71468326,  0.56869015],
          [-0.33875274,  0.46826063],
          [-0.34749244,  0.13441418],
          [ 1.12980796,  0.04281936],
          [-0.38308979,  0.79116812],
          [-0.07425141,  0.2184625 ],
          [-0.44945202, -0.05722266],
          [ 0.85783288,  0.63778888],
          [-0.47486203, -0.22498112],
          [ 0.12627243,  0.86978412],
          [-0.64736458, -0.36342437],
          [ 0.47440459,  1.01101585],
          [-0.38565772, -0.81031183]]),
   array([1, 1, 0, 1, 1, 0, 0, 1, 1, 1, 1, 0, 1, 1, 1, 0, 0, 0, 1, 0, 0, 1,
          1, 1, 0, 0, 1, 0, 0, 0, 1, 1, 1, 1, 1, 0, 0, 1, 0, 0, 1, 1, 1, 0,
          1, 0, 0, 1, 0, 1, 0, 1, 0, 0, 1, 0, 0, 1, 0, 1, 0, 1, 1, 0, 0, 1,
          0, 1, 0, 1, 1, 0, 1, 1, 0, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0,
          1, 1, 0, 0, 1, 1, 0, 1, 0, 1, 0, 0]))),
 (2,
  (array([[-5.66730056,  9.6747529 ],
          [-5.02967294,  8.6596218 ],
          [-6.46936898,  6.82300947],
          [-6.75290119,  7.20976961],
          [-4.14673856,  7.63590025],
          [-5.38927072,  7.95803989],
          [-4.64344634,  7.55830615],
          [-4.46345027,  8.83731915],
          [-4.68748196,  7.21252795],
          [-5.84140615,  8.13223642],
          [-5.36728768,  9.39035577],
          [-5.95944934,  8.31654712],
          [-4.4207936 ,  7.90908652],
          [-6.90411213,  8.20050911],
          [-4.81440963,  8.19155371],
          [-4.95550633,  8.99006291],
          [-4.76708326,  6.78307449],
          [-3.12936539,  7.16255399],
          [-6.43201576,  6.99213819],
          [-7.12970024,  7.7990026 ],
          [-7.61935775,  9.00251464],
          [-4.40874557,  9.27197713],
          [-4.41862884,  6.91289058],
          [-5.8659896 ,  6.93691471],
          [-6.93064951,  8.76263877],
          [-5.56001672,  8.89406765],
          [-3.47162189,  7.76156545],
          [-6.51197348,  8.920786  ],
          [-5.89647284,  7.31403178],
          [-5.04948453,  8.09078804],
          [-5.65507444,  8.71871991],
          [-6.16662202,  7.25381173],
          [-3.73877184,  6.98761474],
          [-7.4653154 ,  9.24937097],
          [-5.9958405 ,  8.38682543],
          [-5.80084772,  6.59052267],
          [-6.43266806,  8.78589697],
          [-5.64701218,  8.97617842],
          [-6.27060303,  7.19945832],
          [-7.22492511,  6.71446709],
          [-7.55929193,  8.67586662],
          [-5.92496887,  5.98552041],
          [-6.37587295,  8.88947751],
          [-7.43787746,  8.95340227],
          [-6.15353935,  8.20288407],
          [-6.54074446,  6.55779297],
          [-4.42173579,  8.81517442],
          [-5.61244473,  7.66386378],
          [-5.81966167,  7.88508551],
          [-4.90351711,  7.53945295],
          [-5.39791168,  8.47356295],
          [-6.86121319,  9.27910763],
          [-6.32089689,  6.7034829 ],
          [-7.64676819,  7.62431237],
          [-5.1184643 ,  9.83531364],
          [-5.36709327,  7.66612429],
          [-6.23402642,  6.12310003],
          [-5.06657892,  7.91513345],
          [-5.56633149,  7.31357851],
          [-5.50099233,  7.05133525],
          [-4.54171545,  8.47599756],
          [-8.20123871,  7.20493971],
          [-6.15849651,  7.17122642],
          [-5.61579958,  9.51942024],
          [-4.76771396,  7.58541058],
          [-6.00874331,  8.14802833],
          [-4.37107585,  7.3410528 ],
          [-6.13764981,  8.56685089],
          [-5.36247649,  8.7494947 ],
          [-5.45052674,  8.99712724],
          [-7.09800301,  9.09917142],
          [-4.90935504,  7.7337076 ],
          [-5.40795882, 10.61018377],
          [-4.41488335,  8.97908848],
          [-5.45939839,  7.7700846 ],
          [-4.93810069, 10.01217222],
          [-6.00556657,  6.93252594],
          [-4.99200386,  7.42740444],
          [-7.07131614,  8.05949363],
          [-5.29052417,  8.70660951],
          [-6.27313606,  7.5595934 ],
          [-4.56369675,  8.12706739],
          [-3.79200136,  9.08201602],
          [-6.71208551, 10.89302579],
          [-7.34687609,  8.35527284],
          [-4.65333349,  7.73756566],
          [-5.63928794,  6.72181978],
          [-5.37253335,  7.08477617],
          [-5.36599885,  7.22067391],
          [-8.41982454,  8.20401253],
          [-3.63234608,  8.33751606],
          [-4.12716808,  9.83742304],
          [-6.75503763,  7.07126671],
          [-5.55684774,  7.30871568],
          [-5.91896553,  8.01811773],
          [-7.14524008,  8.4087608 ],
          [-6.72483849,  6.0175721 ],
          [-6.24123774,  6.95029361],
          [-6.87090971,  6.72508089],
          [-6.27460923,  7.61738757]]),
   array([0, 1, 0, 0, 0, 1, 1, 1, 0, 1, 1, 1, 0, 1, 0, 1, 0, 0, 0, 1, 1, 0,
          0, 0, 1, 1, 0, 1, 0, 1, 1, 1, 0, 1, 0, 0, 1, 0, 0, 0, 1, 1, 1, 0,
          0, 0, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 0, 0, 0, 0, 1, 0, 1, 0, 1,
          0, 0, 0, 0, 1, 1, 1, 1, 0, 1, 0, 0, 0, 0, 1, 0, 1, 1, 1, 1, 0, 0,
          1, 0, 1, 1, 1, 0, 0, 1, 0, 1, 0, 1]))),
 (3,
  (array([[-9.33292655e-01, -1.27122049e+00],
          [-7.66789307e-01,  1.77992147e+00],
          [ 9.03096345e-01,  1.44981703e+00],
          [-1.58918963e+00, -8.32751056e-02],
          [ 9.28016429e-01, -1.23398104e+00],
          [ 1.35985803e+00,  1.51790366e+00],
          [ 5.64146897e-01,  8.34197749e-01],
          [ 3.38467944e-01, -1.54519884e+00],
          [-1.21399342e+00,  4.89499469e-01],
          [-3.51523809e-01,  1.05604000e+00],
          [ 1.42192482e+00,  1.30870092e+00],
          [ 1.18748491e+00,  9.31400904e-01],
          [ 8.19270631e-01,  1.48514069e+00],
          [-1.59251151e+00,  4.85137234e-02],
          [ 1.85219944e-01,  1.53870339e+00],
          [ 1.02943234e+00, -1.81742875e+00],
          [ 1.65055723e+00, -1.19890744e+00],
          [-3.69716837e-01,  1.67597179e+00],
          [-1.77851943e+00,  4.23377975e-01],
          [-4.81676607e-01, -2.04460722e+00],
          [ 2.19893663e+00,  2.99254258e-01],
          [ 4.16115190e-01, -7.85493949e-01],
          [-2.74842332e+00,  2.23314001e-01],
          [ 6.72127536e-01, -1.39095875e+00],
          [ 8.39858504e-01,  1.55542390e+00],
          [ 1.12529264e+00,  2.74200610e-01],
          [-1.22108822e+00, -4.41750034e-01],
          [ 2.78739455e+00, -1.61112841e-03],
          [ 3.53203076e-01,  1.12240129e-01],
          [ 3.10603047e-01, -1.81672854e+00],
          [ 1.86080686e+00, -6.41503310e-01],
          [ 8.94746914e-01, -7.35762193e-01],
          [-1.07707343e+00, -9.91615820e-01],
          [-9.57932339e-01, -9.56397414e-01],
          [-1.96082646e-01, -7.15486741e-01],
          [ 3.23109232e+00, -6.33780279e-01],
          [-1.38817524e+00, -2.48688592e-01],
          [-1.06976424e+00,  9.66055142e-01],
          [ 2.28064080e-01, -1.01741538e+00],
          [ 5.71846228e-01,  1.17039597e+00],
          [-9.47886334e-01, -1.09494698e+00],
          [-2.28879834e+00,  9.91856748e-01],
          [ 2.79871886e-01,  7.12484148e-01],
          [-1.14485872e+00, -7.50568271e-01],
          [-8.08416962e-01, -1.35525390e+00],
          [ 1.15874630e+00,  7.35684201e-02],
          [-1.27191176e+00,  6.74231628e-01],
          [-6.68826500e-01,  1.40599404e+00],
          [-2.33048236e+00,  7.96103665e-01],
          [-1.27620109e+00,  6.92027964e-01],
          [-9.52783841e-01, -1.08481428e+00],
          [ 9.24379467e-02,  1.44550825e+00],
          [ 3.32723963e-01, -1.06577791e+00],
          [-1.02970093e+00, -8.31163579e-01],
          [ 1.30603973e+00,  4.14757047e-01],
          [-1.71750956e+00,  5.00547184e-01],
          [-1.99566063e+00,  1.00972464e+00],
          [ 7.13394018e-01,  1.88985067e+00],
          [-1.79392778e+00,  1.63824076e+00],
          [ 5.19077120e-01, -5.54142502e-01],
          [ 1.41713525e+00, -1.13359774e+00],
          [-8.48825636e-01, -1.14486125e+00],
          [ 1.81154534e+00,  4.97164230e-01],
          [-2.14823656e+00,  7.33412925e-01],
          [ 1.06731305e+00,  1.63059220e+00],
          [-1.73590855e+00,  3.20189254e-01],
          [ 1.02467738e+00, -1.25475559e+00],
          [ 2.21517183e+00, -5.96284790e-01],
          [ 1.14752935e+00,  8.57570034e-01],
          [-1.55245701e+00, -3.36171152e-02],
          [ 2.55716114e+00, -2.99789421e-01],
          [ 7.94490859e-01,  8.74280300e-01],
          [-1.16336461e+00,  2.32337605e-01],
          [-3.94536291e-01,  1.19313340e+00],
          [-1.13456945e+00, -7.49047339e-01],
          [ 1.00666441e+00, -1.55511830e+00],
          [-2.34109735e+00,  4.52143651e-01],
          [-7.90107843e-01, -1.40539861e+00],
          [-2.49600849e-02,  1.01556610e+00],
          [ 1.88879091e+00, -1.27159355e+00],
          [ 2.01763813e+00, -6.80769118e-01],
          [-7.21544129e-01, -1.56319001e+00],
          [ 3.28795018e-01,  9.88507261e-01],
          [-2.39486480e+00,  8.87560308e-01],
          [-9.37827394e-01,  8.55915239e-01],
          [ 5.93709424e-01, -8.83762427e-01],
          [ 2.68917440e-01,  1.76442408e+00],
          [-1.61918481e+00,  4.36660291e-01],
          [-2.25592991e-01, -2.72499883e+00],
          [ 2.81655475e-01,  6.83987133e-01],
          [-8.51055761e-01, -1.22979364e+00],
          [ 1.74848466e+00,  1.17389035e+00],
          [ 1.12440472e+00,  5.85181570e-01],
          [-5.00014133e-01,  1.58890117e+00],
          [ 7.13425522e-01,  1.11017445e+00],
          [ 5.70065432e-01,  1.58754104e+00],
          [-1.10825112e+00, -5.84147570e-01],
          [-1.04162242e+00, -8.80162451e-01],
          [ 7.51042801e-01, -1.50398321e+00],
          [ 8.50404935e-01, -7.86309791e-01]]),
   array([0, 0, 1, 0, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0, 1, 1, 1, 0, 0, 0, 1, 1,
          0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1, 0, 0, 1, 0,
          0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 1, 0, 1, 1, 0, 1, 0, 1, 0,
          1, 1, 1, 0, 1, 1, 0, 0, 0, 1, 0, 0, 0, 1, 1, 0, 1, 0, 0, 1, 0, 0,
          0, 1, 0, 1, 1, 0, 1, 0, 0, 1, 1, 1])))]

nrows=len(datasets)
ncols=len(Kernel) + 1

fig, axes = plt.subplots(nrows, ncols,figsize=(20,16))

#第一层循环：在不同的数据集中循环
for ds_cnt, (X,Y) in enumerate(datasets):

    #在图像中的第一列，放置原数据的分布
    ax = axes[ds_cnt, 0]
    if ds_cnt == 0:
        ax.set_title("Input data")
    ax.scatter(X[:, 0], X[:, 1], c=Y, zorder=10, cmap=plt.cm.Paired,edgecolors='k') 
    ax.set_xticks(())
    ax.set_yticks(())

    #第二层循环：在不同的核函数中循环
    #从图像的第二列开始，一个个填充分类结果
    for est_idx, kernel in enumerate(Kernel):

        #定义子图位置
        ax = axes[ds_cnt, est_idx + 1]

        #建模
        clf = svm.SVC(kernel=kernel, gamma=2).fit(X, Y)
        score = clf.score(X, Y)

        #绘制图像本身分布的散点图
        ax.scatter(X[:, 0], X[:, 1], c=Y
                ,zorder=10
                ,cmap=plt.cm.Paired,edgecolors='k')
        #绘制支持向量
        ax.scatter(clf.support_vectors_[:, 0], clf.support_vectors_[:, 1], s=50, facecolors='none', zorder=10, edgecolors='k')
        #绘制决策边界
        x_min, x_max = X[:, 0].min() - .5, X[:, 0].max() + .5
        y_min, y_max = X[:, 1].min() - .5, X[:, 1].max() + .5

        #np.mgrid，合并了我们之前使用的np.linspace和np.meshgrid的用法
        #一次性使用最大值和最小值来生成网格
        #表示为[起始值：结束值：步长]
        #如果步长是复数，则其整数部分就是起始值和结束值之间创建的点的数量，并且结束值被包含在内
        XX, YY = np.mgrid[x_min:x_max:200j, y_min:y_max:200j]
        #np.c_，类似于np.vstack的功能
        Z = clf.decision_function(np.c_[XX.ravel(), YY.ravel()]).reshape(XX.shape)
        #填充等高线不同区域的颜色
        ax.pcolormesh(XX, YY, Z > 0, cmap=plt.cm.Paired)
        #绘制等高线
        ax.contour(XX, YY, Z, colors=['k', 'k', 'k'], linestyles=['--', '-', '--'], levels=[-1, 0, 1])
        #设定坐标轴为不显示
        ax.set_xticks(())
        ax.set_yticks(())

        #将标题放在第一行的顶上
        if ds_cnt == 0:
            ax.set_title(kernel)

        #为每张图添加分类的分数
        ax.text(0.95, 0.06, ('%.2f' % score).lstrip('0')
                    , size=15
                    , bbox=dict(boxstyle='round', alpha=0.8, facecolor='white')
                    #为分数添加一个白色的格子作为底色
                    , transform=ax.transAxes #确定文字所对应的坐标轴，就是ax子图的坐标轴本身
                    , horizontalalignment='right' #位于坐标轴的什么方向
                    )

plt.tight_layout()
plt.show()

结果：

可以观察到，线性核函数和多项式核函数在非线性数据上表现会浮动，如果数据相对线性可分，则表现不错，如果是像环形数据那样彻底不可分的，则表现糟糕。在线性数据集上，线性核函数和多项式核函数即便有扰动项也可以表现不错，可见多项式核函数是虽然也可以处理非线性情况，但更偏向于线性的功能。

Sigmoid核函数就比较尴尬了，它在非线性数据上强于两个线性核函数，但效果明显不如rbf，它在线性数据上完全比不上线性的核函数们，对扰动项的抵抗也比较弱，所以它功能比较弱小，很少被用到。

rbf，高斯径向基核函数基本在任何数据集上都表现不错，属于比较万能的核函数。我个人的经验是，无论如何先试试看高斯径向基核函数，它适用于核转换到很高的空间的情况，在各种情况下往往效果都很不错，如果rbf效果不好，那我们再试试看其他的核函数。另外，多项式核函数多被用于图像处理之中。

Grateful_Dead424

关注

0
点赞
踩
2

收藏

觉得还不错? 一键收藏
0
评论
Sklearn(v3)——SVM理论(2)

from sklearn.datasets import make_blobsfrom sklearn.svm import SVCimport matplotlib.pyplot as pltimport numpy as npX,y = make_blobs(n_samples=50,centers=2,random_state=0,cluster_std=0.6)plt.scatter(X[:,0],X[:,1],c=y,s=50,cmap="rainbow")plt.xticks(.
复制链接

扫一扫