头歌实训-机器学习（逻辑回归）

1.逻辑回归简述

#encoding=utf8
import numpy as np

def sigmoid(t):
    '''
    完成sigmoid函数计算
    :param t: 负无穷到正无穷的实数
    :return: 转换后的概率值
    :可以考虑使用np.exp()函数
    '''
    #********** Begin **********#
    return  1.0/(1+np.exp(-t))
    #********** End **********#

2.逻辑回归算法详解

from sklearn import datasets
from sklearn.datasets import load_iris 
import numpy as np
import math
from sklearn.model_selection import train_test_split
from collections import Counter
from sklearn.linear_model import LogisticRegression   #导入逻辑回归模型

#########Begin########
# 导入数据
iris = datasets.load_iris()
X= iris['data']
y = iris['target']
X = X[y!=2] #  筛选数据，只选择标签为0和1
y=y[y!=2]
# 数据划分
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)  

# 模型调用
model = LogisticRegression()  
# 模型训练
model.fit(X_train, y_train) 
# 数据预测
y_pred = model.predict(X_test)  
# 结果打印
print("准确度:",model.score(X_test,y_test))

########End#########

3.sklearn逻辑回归 - 手写数字识别

from sklearn.linear_model import LogisticRegression

def digit_predict(train_image, train_label, test_image):
    '''
    实现功能：训练模型并输出预测结果
    :param train_sample: 包含多条训练样本的样本集，类型为ndarray,shape为[-1, 8, 8]
    :param train_label: 包含多条训练样本标签的标签集，类型为ndarray
    :param test_sample: 包含多条测试样本的测试集，类型为ndarry
    :return: test_sample对应的预测标签
    '''

    #************* Begin ************#
    # 训练集变形
    flat_train_image = train_image.reshape((-1, 64))
    # 训练集标准化
    train_min = flat_train_image.min()
    train_max = flat_train_image.max()
    flat_train_image = (flat_train_image-train_min)/(train_max-train_min)
    # 测试集变形
    flat_test_image = test_image.reshape((-1, 64))
    # 测试集标准化
    test_min = flat_test_image.min()
    test_max = flat_test_image.max()
    flat_test_image = (flat_test_image - test_min) / (test_max - test_min)

    # 训练--预测
    rf = LogisticRegression(C=4.0)
    rf.fit(flat_train_image, train_label)
    return rf.predict(flat_test_image)

    #************* End **************#

4.逻辑回归案例 - 癌细胞精准识别

# -*- coding: utf-8 -*-

import numpy as np
import warnings
warnings.filterwarnings("ignore")

def sigmoid(x):
    '''
    sigmoid函数
    :param x: 转换前的输入
    :return: 转换后的概率
    '''
    return 1/(1+np.exp(-x))


def fit(x,y,eta=1e-3,n_iters=10000):
    '''
    训练逻辑回归模型
    :param x: 训练集特征数据，类型为ndarray
    :param y: 训练集标签，类型为ndarray
    :param eta: 学习率，类型为float
    :param n_iters: 训练轮数，类型为int
    :return: 模型参数，类型为ndarray
    '''
    #   请在此添加实现代码   #
    #********** Begin *********#
    theta = np.zeros(x.shape[1])
    i_iter = 0
    while i_iter < n_iters:
        gradient = (sigmoid(x.dot(theta))-y).dot(x)
        theta = theta -eta*gradient
        i_iter += 1
    return theta

    #********** End **********#