逻辑回归的python实现

最新推荐文章于 2022-04-30 12:34:24 发布

小镇大爱

最新推荐文章于 2022-04-30 12:34:24 发布

阅读量213

点赞数

分类专栏： machine learning

本文链接：https://blog.csdn.net/qq_38343151/article/details/106747408

版权

machine learning 专栏收录该内容

3 篇文章 0 订阅

订阅专栏

梯度下降方法和scikit-learn两种方法预测学生是否被录取，需要数据集的可留言

##方法1
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

def sigmoid(z):
    return 1/(1+np.exp(-z))
def predict(theta,X):
    prob = sigmoid(X*theta.T)
    return [1 if a>=0.5 else 0 for a in prob]
def gradientDescent(X,y,theta,alpha,m,numIter):
    XTrans = X.transpose

    t = theta  

    for i in range(0,numIter):
        gradient=0
        theta = np.mat(t)
        pred = np.array(predict(theta,X))
        loss = pred - y
        gradient = np.dot(XTrans,loss)
        theta = theta-alpha*gradient ###参数更新
    return theta

##开始预测，对学生是否被录取的逻辑回归预测
df = pd.read_csv('./data/5_logisitic_admit.csv')
df.insert(1,'Ones',1)
#print(df.head(10))
positive = df[df['admit']==1]
negative = df[df['admit']==0]
fig,ax = plt.subplots(figsize=(8,5))
ax.scatter(positive['gre'],positive['gpa'],s=30,c='b',marker='o',label='admit')
ax.scatter(negative['gre'],negative['gpa'],s=30,c='r',marker='x',label='not admit')
ax.legend()
ax.set_xlabel('gre')
ax.set_ylabel('gpa')
#plt.show()
x = df.iloc[:,1:4]
y = df['admit']
x = np.array(x.values)
y = np.array(y.values)
m,n = np.shape(x)
print(m,n)
t = np.ones(n)
numIter = 1000
alpha = 0.00001
theta = gradientDescent(x,y,t,alpha,m,numIter)
pred = Predict(theta,x)

##方法2
from sklearn.linear_model.logistic import LogisticRegression
lf = LogisticRegression()
lf.fit(x,y)
pred=lf.predict(x)
print(pred)