01-赵志勇机器学习-Logistics_Regression-train

Logistics Regression

二分类问题。

模型	线性模型
响应	sigmoid
损失函数（显示）	最小均方
优化方法	BGD

例子：

#coding utf-8
import numpy as np

def load_data(file_name):
  feature_data = []
  label_data = []

  f = open(file_name) # 打开文件
  for line in f.readlines():
    # @ strip() 去除字符串首尾的空格
	# @ split("\t") 以“\t”分割字符串
    lines = line.strip().split("\t")

    feature_tmp = []
    label_tmp = []
    feature_tmp.append(1) # 偏置项

    for i in range(len(lines)-1):
      feature_tmp.append(float(lines[i]))
    label_tmp.append(float(lines[-1]))

    feature_data.append(feature_tmp)
    label_data.append(label_tmp)

  f.close() # 关闭文件

  return np.mat(feature_data), np.mat(label_data)

def sig(x):
  return 1.0/(1+np.exp(-x))

def compute_error(h, label):
  # @ shape() 获得特征的长度，[0]行数，[1]列数
  n = np.shape(h)[0]

  err = 0
  for i in range(n):
    if h[i,0]>0 and (1-h[i,0])>0:
      err -= (label[i,0]*np.log(h[i,0])
	      + (1-label[i,0])*np.log(1-h[i,0]))
    else:
      err -= 0

  return err

def lr_train_bgd(feature, label, maxCycle, alpha):
  n = np.shape(feature)[1]
  W = np.mat(np.ones((n,1)))

  for i in range(maxCycle):
    h = sig(feature*W)
    err = label - h
    if i % 100 == 0:
      print(compute_error(h, label))

    W = W + alpha * feature.T * err

  return W

def save_model(file_name, W):
  f = open(file_name, "w")
  w_array = []
  n = np.shape(W)[0]
  for i in range(n):
    w_array.append(str(W[i,0]))

  f.write("\t".join(w_array))
  f.close()

if __name__ == "__main__":
  print("load data")
  feature, label = load_data("data.txt")
  print("train")
  w = lr_train_bgd(feature, label, 1000, 0.1)
  print("save")
  save_model("weights2018", w)

　　

参考：

https://blog.csdn.net/google19890102/article/details/77996085

https://blog.csdn.net/google19890102?viewmode=contents

https://github.com/zhaozhiyong19890102/Python-Machine-Learning-Algorithm