用的数据集
import numpy as np import pandas as pd import numpy.random import time import matplotlib.pyplot as plt %matplotlib inline data=pd.read_csv('Folds5x2_pp.CSV')就是开始梯度大大大大大大，当时就觉得不太对，但是懒得重新看了，给数据处理下凑合用吧
a1=data.loc[:,"AT"].max() a2=data.loc[:,"V"].max() a3=data.loc[:,"AP"].max() a4=data.loc[:,"RH"].max() a5=data.loc[:,"PE"].max() data=pd.DataFrame({'a1':data['AT']/a1, 'a2':data['V']/a2, 'a3':data['AP']/a3, 'a4':data['RH']/a4, 'a5':data['PE']/a5}) data.insert(0,'f',1) Udata=data.values x=Udata[:,0:5] y=Udata[:,5:6] theta=np.zeros([1,5]) def hx(x,theta): return np.dot(x,theta.T) def loss(x,y,theta): a=np.square(hx(x,theta)-y) return a.sum()/(len(y)) def pd(x,y,theta): grad=np.zeros(theta.shape)#梯度结果，占位 error=(y-hx(x,theta)).ravel()# ravel 多维数组拉成一维数组 for i in range(len(theta.ravel())):# 所有参数的偏导 grad[0,i]=(np.sum(np.multiply(error,x[:,i]))) return grad def DM2(x,y,theta,YuZhi,learnning_Rate): start_time=time.time() i=0 k=0 grad=np.zeros(theta.shape) Loss=[loss(x,y,theta)] while True: grad=pd(x[k:k+1],y[k:k+1],theta) k+=1 if k>=len(x): k=0 theta=theta+(learnning_Rate*grad) Loss.append(loss(x,y,theta)) i+=1 if i>YuZhi: break return theta,i-1,Loss,time.time()-start_time ENDtheta,times,ENDcosts,spend=DM2(x,y,theta,15000,0.0001) print('耗时：%f'%(spend)) fig,ax=plt.subplots(figsize=(12,4)) ax.plot(np.arange(len(ENDcosts)),ENDcosts,'r') ax.set_xlabel("Iter") ax.set_ylabel("Loss") 耗时：2.377032 ENDcosts[-1] 0.006015465001446885