Python3入门机器学习 - 逻辑回归与决策边界

2021-10-20 01:08:32

logistic回归又称logistic回归分析，是一种广义的线性回归分析模型,以胃癌病情分析为例，选择两组人群，一组是胃癌组，一组是非胃癌组，两组人群必定具有不同的体征与生活方式等。因此因变量就为是否胃癌，值为“是”或“否”，自变量就可以包括很多了，如年龄、性别、饮食习惯、幽门螺杆菌感染等。

逻辑回归需要将原本线性回归结果的值域置于（0，1）之间，概率大于0.5看作结果为1

常使用sigmoid函数将结果变为（0，1）之间的值域

绘制sigmoid曲线

def sigmoid(t):
    return 1/(1+np.exp(-t))

x = np.linspace(-10,10,100)
y = sigmoid(x)
plt.plot(x,y)

逻辑回归的损失函数

该损失函数没有公式解，可以用梯度下降法求最优解

损失函数求导

封装Logistics模型

我使用Logistics模型及鸢尾花数据集的前两列，可得到1.0准确率的预测精准度。

# 代码与线性回归及其相似，只是推导公式不同
# _*_ encoding:utf-8 _*_
import numpy as np
from sklearn.metrics import r2_score
from metrics import accuracy_score

class LinearRegression:
    def __init__(self):
        self.coef_ = None
        self.interception_ = None
        self._theta = None

    def _sigmoid(self,t):
        return 1./(1.+ np.exp(-t))

    def fit(self,X_train,y_train,eta=0.01,n_iters=1e6):
        def J(theta,X_b,y):
            y_hat = self._sigmoid(X_b.dot(theta))
            try:
                return -np.sum(y*np.log(y_hat)+(1-y)*np.log(1-y_hat))/len(y)
            except:
                return float("inf")
        def dJ(theta,X_b,y):
            # res = np.empty()
            # res[0] = np.sum(X_b.dot(theta)-y)
            # for i in range(1,len(theta)):
            #     res[i] = (X_b.dot(theta)-y).dot(X_b[:,i])
            # return res * 2 / len(X_b)
            return X_b.T.dot(self._sigmoid(X_b.dot(theta))-y)/len(X_b)
        def gradient_descent(X_b,y,initial_theta,eta,n_iters=1e6,epsilon=1e-8):
            theta = initial_theta
            cur_iter = 0
            while cur_iter<n_iters:
                gradient = dJ(theta,X_b,y)
                last_theta = theta
                theta = theta - eta * gradient
                if (abs(J(theta,X_b,y) - J(last_theta,X_b,y)) < epsilon):
                    break
                cur_iter+=1
            return theta
        X_b = np.hstack([np.ones((len(X_train),1)),X_train])
        initial_theta = np.zeros(X_b.shape[1])
        self._theta = gradient_descent(X_b,y_train,initial_theta,eta,n_iters)
        self.interception_ = self._theta[0]
        self.coef_ = self._theta[1:
        return self


    def predict_proba(self,X_predict):
        X_b = np.hstack([np.ones((len(X_predict),1)),X_predict])
        return self._sigmoid(X_b.dot(self._theta))

    def predict(self,X_predict):
        proba = self.predict_proba(X_predict)
        return proba>=0.5

    def score(self,X_test,y_test):
        return accuracy_score(y_test,self.predict(X_test))

    def __repr__(self):
        return "LogisticRegreesion()"

决策边界

在逻辑回归中，易得决策边界为 theta*X_b=0 的直线

如果只有两个特征值，则很容易通过公式画出逻辑回归的决策边界

蓝线即为鸢尾花数据前两列特征值通过逻辑回归得到的决策边界

不规则的决策边界的绘制

一种绘制思路

def plot_decision_boundary(model,axis):
    x0,x1 = np.meshgrid(
        np.linspace(axis[0],axis[1],int((axis[1]-axis[0])*100)),
        np.linspace(axis[2],axis[3],int((axis[3]-axis[2])*100))
    )
    X_new = np.c_[x0.ravel(),x1.ravel()]
    
    y_predict = model.predict(X_new)
    zz = y_predict.reshape(x0.shape)
    
    from matplotlib.colors import ListedColormap
    custom_cmap = ListedColormap(['#EF9A9A','#FFF59D','#90CAF9'])
    
    plt.contourf(x0,x1,zz,linewidth=5,cmap=custom_cmap)

plot_decision_boundary(log_reg,axis=[4,7.5,1.5,4.5])
plt.scatter(X[y==0,0],X[y==0,1])
plt.scatter(X[y==1,0],X[y==1,1])
plt.show()

knn算法的决策边界

knn_clf = KNeighborsClassifier()
knn_clf.fit(iris.data[:,:2],iris.target)

plot_decision_boundary(knn_clf,axis=[4,7.5,1.5,4.5])
plt.scatter(X[y==0,0],X[y==0,1])
plt.scatter(X[y==1,0],X[y==1,1])
plt.show()

n_neighbors默认等于5时

n_neighbors等于50时

多项式特征应用于逻辑回归

#准备数据
X = np.random.normal(0,1,size=(200,2))
y = np.array(X[:,0]**2 + X[:,1]**2<1.5,dtype='int')

def PolynomialLogisticRegression(degree):
    return Pipeline([
        ('Poly',PolynomialFeatures(degree=degree)),
        ('std_scaler',StandardScaler()),
        ('Logistic',LogisticRegression())
    ])

Log_reg = PolynomialLogisticRegression(2)

Log_reg.fit(X,y)

逻辑回归的模型正则化

逻辑回归的模型正则化方式

#准备数据
import numpy as np
import matplotlib.pyplot as plt

X = np.random.normal(0,1,size=(200,2))
y = np.array(X[:,0]**2 + X[:,1]<1.5,dtype='int')

for _ in range(20):
    y[np.random.randint(200)] = 1   #噪音

plt .scatter(X[y==0,0],X[y==0,1])
plt .scatter(X[y==1,0],X[y==1,1])
plt.show()

数据.png

from sklearn.linear_model import LogisticRegression
from sklearn.preprocessing import PolynomialFeatures
from sklearn.pipeline import Pipeline

log_reg = LogisticRegression()

log_reg.fit(X,y)

LogisticRegression(C=1.0, class_weight=None, dual=False, fit_intercept=True,
intercept_scaling=1, max_iter=100, multi_class='ovr', n_jobs=1,
penalty='l2', random_state=None, solver='liblinear', tol=0.0001,
verbose=0, warm_start=False)

其中模型正则化公式中的参数C默认为1,penalty默认为'l2'

from sklearn.preprocessing import StandardScaler

def PolynomialLogisticRegression(degree,C=1.0,penalty='l2'):
    return Pipeline([
        ('Poly',PolynomialFeatures(degree=degree)),
        ('std_scaler',StandardScaler()),
        ('Logistic',LogisticRegression(C=C,penalty=penalty))
    ])

poly_log_reg = PolynomialLogisticRegression(degree=20,C=0.1,penalty='l1')
poly_log_reg.fit(X,y)

曲线相对平滑

应用OVR和OVO使逻辑回归处理多分类问题

OVR:One Vs Rest
OVO:One Vs One
OVR耗时较少，性能较高，但分类准确度略低
OVO耗时较多，分类准确度较高

#为了数据可视化方便，我们只使用鸢尾花数据集的前两列特征
from sklearn import datasets

iris = datasets.load_iris()
X = iris['data'][:,:2]
y = iris['target']

log_reg = LogisticRegression(multi_class='ovr')   #传入multi_class参数可以指定使用ovr或ovo，默认ovr
log_reg.score(X_test,y_test)
>>> 0.578   #由于只使用前两列特征，导致分类准确度较低
log_reg = LogisticRegression(multi_class='ovr',solver='newton-cg')
log_reg.fit(X_train,y_train)
log_reg.score(X_test,y_test)
>>> 0.7894736842105263

OVR分类决策边界

OVO分类决策边界

使用scikitlearn中的OVO及OVR类来进行多分类

from sklearn.multiclass import OneVsOneClassifier
from sklearn.multiclass import OneVsRestClassifier

ovr = OneVsRestClassifier(log_reg)
ovr.fit(X_train,y_train)
print(ovr.score(X_test,y_test))

ovo = OneVsOneClassifier(log_reg)
ovo.fit(X_train,y_train)
print(ovo.score(X_test,y_test))

>>> 0.7894736842105263
>>> 0.8157894736842105

码农公寓