import numpy as np#1采樣數據data = []#保存樣本集的列表for i in range(100):#循環采樣100個點 x = np.random.uniform(-10,10)#隨機采樣輸入X # 采樣高斯噪聲 eps = np.random.normal(0.,0.1) #得到模型的輸出 y = 1.477*x +0.089 +eps data.append([x,y])#保存樣本點data = np.array(data)#轉換為2維數組print(data)#2計算誤差def mse(b,w,points): #根據當前的w，b參數計算軍方差損失 totalError = 0 for i in range(0,len(points)):#循環迭代所有點 x = points[i,0]# 獲得i號店的輸入x y = points[i,1]# 獲得i號點的輸出y totalError+=(y-(w*x+b))**2 return totalError/float(len(points))#得到均方差#3，計算梯度def step_gradient(b_current,w_current,points,lr): #計算誤差函數在所有點上的導數，并更新w，b b_gradient = 0 w_gradient =0 M = float(len(points)) for i in range(0,len(points)): x = points[i,0] y=points[i,1] b_gradient +=(2/M)*((w_current*x+b_current)-y) w_gradient += (2/M) * x*((w_current*x+b_current) - y) #根據梯度下降算法更新w,b，lr為xuexilv new_b = b_current -(lr*b_gradient) new_w = w_current -(lr * w_gradient) return [new_b,new_w]#梯度更新def gradient_descent(points,starting_b,starting_w,lr,num_iterations): b = starting_b w = starting_w for step in range(num_iterations): b,w = step_gradient(b,w,np.array(points),lr) loss = mse(b,w,points) if step%50 == 0: print("iteration:{},loss:{},w:{},b:{}".format(step,loss,w,b)) return [b,w]def main(): lr = 0.01 initial_b=0 initial_w = 0 num_iterations = 1000 [b,w] = gradient_descent(data,initial_b,initial_w,lr,num_iterations) loss = mse(b,w,data) print("final loss:{},w:{},b:{}".format(loss,w,b))
