【感知机模型】手写代码训练 / 使用sklearn的Perceptron模块训练

读取原始数据

import pandas as pd
import numpy as np
in_data = pd.read_table('./origin-data/perceptron_15.dat', sep='s+', header=None)
X_train = np.array(in_data.loc[:,[0,1,2,3]])
y_train = np.array(in_data[4])

训练感知机模型

class MyPerceptron:
  def __init__(self):
    self.w = None
    self.b = 0
    self.l_rate = 1

  def fit(self, X_train, y_train):
  #用样本点的特征数更新初始w,如x1=(3,3)T,有两个特征,则self.w=[0,0]
    self.w = np.zeros(X_train.shape[1])
    i = 0
    while i < X_train.shape[0]:
      X = X_train[i]
      y = y_train[i]
      # 如果y*(wx+b)≤0 说明是误判点,更新w,b
      if y * (np.dot(self.w, X) + self.b) <= 0:
        self.w += self.l_rate * np.dot(y, X)
        self.b += self.l_rate * y
        i=0 #如果是误判点,从头进行检测
      else:
        i+=1
from sklearn.linear_model import Perceptron

# 使用sklearn中的Perceptron类训练
perceptron = Perceptron()
time1 = datetime.datetime.now()
perceptron.fit(X_train, y_train)
time2 = datetime.datetime.now()
print("共用时:", (time2-time1).microseconds, "微秒")
print(perceptron.coef_)
print(perceptron.intercept_)

共用时: 4769 微秒
[[ 2.9686576 -1.513057 2.211151 4.227677 ]]
[-3.]

# 使用自己写的MyPerceptron类训练
perceptron = MyPerceptron()
time1 = datetime.datetime.now()
perceptron.fit(X_train, y_train)
time2 = datetime.datetime.now()
print("共用时:", (time2-time1).microseconds, "微秒")
print(perceptron.w)
print(perceptron.b)

共用时: 12479 微秒
[ 3.6161856 -2.013502 3.123158 5.49830856]
-4

原文地址:https://www.cnblogs.com/yanqiang/p/12038266.html