import matplotlib.pyplot as plt import numpy as np import pandas as pdnp.set_printoptions(threshold=np.inf) # 解決顯示不完全問題 plt.rcParams['font.sans-serif'] = ['Microsoft YaHei'] plt.rcParams['axes.unicode_minus'] = Falsedef load_data():datafile = "F:\PyCharm\PyCharm文件\波士頓房價預測\housing.data"data = np.fromfile(datafile, sep=' ')feature_names = ['CRIM', 'ZN', 'INDUS', 'CHAS', 'NOX', 'RM', 'AGE', 'DIS', 'RAD', 'TAX', 'PTRATIO', 'B', 'LSTATA','MEDV']feature_num = len(feature_names) # 特征值數量data = data.reshape([data.shape[0] // feature_num, feature_num])ratio = 0.8offset = int(data.shape[0] * ratio)training_data = data[:offset]maximums, minimums, avgs = training_data.max(axis=0), training_data.min(axis=0), training_data.sum(axis=0) / \training_data.shape[0]for i in range(feature_num):data[:, i] = (data[:, i] - minimums[i]) / (maximums[i] - minimums[i])training_data = data[:offset]test_data = data[offset:]return training_data, test_datatraining_data, test_data = load_data() x = training_data[:, :-1] y = training_data[:, -1:]class Network(object):def __init__(self, num_of_weights): # num_of_weights=13np.random.seed(0)self.w = np.random.randn(num_of_weights, 1)self.b = 0def forward(self, x):z = np.dot(x, self.w) + self.b # 404*1return zdef loss(self, z, y):error = z - ynum_samples = error.shape[0] # 404*1cost = error * error # 404*1cost = np.sum(cost) / num_samplesreturn costdef gradient(self, x, y, z):gradient_w = (z - y) * x # (z-y):404*1,gradient_w:404*12gradient_w = np.mean(gradient_w, axis=0) # 計算504*13各列平均值，得到13*1的矩陣gradient_w = gradient_w[:, np.newaxis] # 轉置13*1gradient_b = (z - y) # 404*1gradient_b = np.mean(gradient_b) # 一個數return gradient_w, gradient_b # 最后存在13個w，一個bdef update(self, gradient_w, gradient_b, eta=0.01): # eta：設置步長又稱學習率self.w = self.w - eta * gradient_w # 對參數w進行微調self.b = self.b - eta * gradient_b # 對參數b進行微調def train(self, x, y, iterations=100, eta=0.01):losses = []for i in range(iterations): # 循環1000次z = self.forward(x) # 進行前向運算L = self.loss(z, y) # 計算損失值404*1gradient_w, gradient_b = self.gradient(x, y, z) # 計算梯度self.update(gradient_w, gradient_b, eta) # 更新梯度，進行微調losses.append(L)if (i + 1) % 10 == 0:print('iter{},loss{}'.format(i, L)) # 每十輪輸出一波損失值print(self.w) # 輸出各個變量權重print(self.b) # 輸出偏移矢量return losses # 返回iteration次每次的損失值net = Network(13) # wx+b b是12個變量共用一個 num_iterations = 1000 # 迭代次數 losses = net.train(x, y, iterations=num_iterations, eta=0.01) plot_x = np.arange(num_iterations) plot_y = np.array(losses) plt.plot(plot_x, plot_y) plt.title("損失值") plt.show()

總結

以上是生活随笔為你收集整理的波士顿房价预测模型源代码的全部內容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。