感知器代码

'''******************导入数据包*********************************'''

import pandas as pd 
import numpy as np
import random
from sklearn import preprocessing

'''***************导入数据与数据处理*****************************'''

address = 'F:/python code/network_perceotron/basedata.csv' 
file = pd.read_csv(address,encoding='GBK',header=None)
data = np.array(file)
data = data[1:9,1:6]
data = np.array(data,dtype=float)
add = np.ones(8)
x_train = np.column_stack((data,add))
x_train = preprocessing.scale(x_train)

'''********************封装感知器*********************************'''

class perceptron:

    #   初始化参数
    def __init__(self):
        self.X = x_train
        self.W = np.zeros(6).reshape(6,1)

    #   预测
    def predict(self,W,X):
        z = np.tanh(np.dot(X,W))
        return z

    #   计算误差
    def loss(self,label,z):
        label = y_train
        output = z
        error = label - output
        return error

    #   更新权重
    def update(self,W,X,error):
        lr = np.random.rand()
        self.W  += lr*(np.dot(X.T,error))
        return W

'''***************************训练数据***************************************'''
net = perceptron()
W = net.W
X = net.X
y_train = np.array([-1,-1,-1,-1,1,1,1,1]).reshape(8,1)
itr_num = int(input("请输入迭代次数："))

for i in range(itr_num):
    #   预测
    y = net.predict(W,X)
    #   计算损失
    los = net.loss(y_train,y)
    #   更新权重
    W = net.update(W,X,los)

'''**************************——————测试————**********************************'''

#   输入测试数据集
X_test = np.array([[0.052,0.084,0.021,0.037,0.022,1],
                   [0.074,0.083,0.105,0.19,1,1]])
#   预处理数据
X_test = preprocessing.scale(X_test)
W_best = W
prdit_y = net.predict(W_best,X_test)
print(prdit_y)

数据集