原文链接：https://blog.csdn.net/Mind_programmonkey/article/details/89606115
数据处理
import os
import numpy as np
import pandas as pd
# 定义标准化的函数
def normalize_feature(df):
return df.apply(lambda column:(column-column.mean())/column.std())
# 读取csv
csv_data = normalize_feature(pd.read_csv("F:/NoteBookProject/Demo2.csv"))
print(csv_data.head())
ones = pd.DataFrame({"ones": np.ones(len(csv_data))})
csv_data = pd.concat([ones,csv_data],axis=1)
print(csv_data.head())
X_data = np.array(csv_data[csv_data.columns[0:3]])
y_data = np.array(csv_data[csv_data.columns[-1]]).reshape(len(csv_data),1)
print(X_data.shape, type(X_data))
print(y_data.shape, type(y_data))
输出
square bedrooms price
0 0.130010 -0.223675 0.475747
1 -0.504190 -0.223675 -0.084074
2 0.502476 -0.223675 0.228626
3 -0.735723 -1.537767 -0.867025
4 1.257476 1.090417 1.595389
ones square bedrooms price
0 1.0 0.130010 -0.223675 0.475747
1 1.0 -0.504190 -0.223675 -0.084074
2 1.0 0.502476 -0.223675 0.228626
3 1.0 -0.735723 -1.537767 -0.867025
4 1.0 1.257476 1.090417 1.595389
(47, 3) <class 'numpy.ndarray'>
(47, 1) <class 'numpy.ndarray'>
训练
import tensorflow as tf
alpha = 0.01 # 学习率 alpha
epoch = 500 # 训练全量数据集的轮数
with tf.name_scope('input'):
# 输入 X，形状[47, 3]
X = tf.placeholder(tf.float32, X_data.shape, name='X')
# 输出 y，形状[47, 1]
y = tf.placeholder(tf.float32, y_data.shape, name='y')
with tf.name_scope('hypothesis'):
# 权重变量 W，形状[3,1]
W = tf.get_variable("weights", (X_data.shape[1], 1), initializer=tf.constant_initializer())
# 假设函数 h(x) = w0*x0+w1*x1+w2*x2, 其中x0恒为1代表bias
# 推理值 y_pred 形状[47,1]
y_pred = tf.matmul(X, W, name='y_pred')
with tf.name_scope('loss'):
# 损失函数采用最小二乘法，y_pred - y 是形如[47, 1]的向量。
# tf.matmul(a,b,transpose_a=True)，transpose_a=True矩阵a进行转置 表示：矩阵a的转置乘矩阵b，即 [1,47] X [47,1]
# 损失函数操作 loss
loss_op = 1 / (2 * len(X_data)) * tf.matmul((y_pred - y), (y_pred - y), transpose_a=True)
with tf.name_scope('train'):
# 随机梯度下降优化器 opt
train_op = tf.train.GradientDescentOptimizer(learning_rate=alpha).minimize(loss_op)
with tf.Session() as sess:
# 初始化全局变量
sess.run(tf.global_variables_initializer())
# 开始训练模型
# 因为训练集较小，所以采用批梯度下降优化算法，每次都使用全量数据训练
for e in range(1, epoch + 1):
_, loss, w = sess.run([train_op, loss_op, W], feed_dict={X: X_data, y: y_data})
if e % 10 == 0:
log_str = "Epoch %d \t Loss=%.4g \t Model: y = %.4gx1 + %.4gx2 + %.4g"
print(log_str % (e, loss, w[1], w[2], w[0]))
输出
Epoch 10 Loss=0.4184 Model: y = 0.0791x1 + 0.03948x2 + 3.353e-10
Epoch 20 Loss=0.3582 Model: y = 0.1489x1 + 0.07135x2 + -5.588e-11
Epoch 30 Loss=0.3126 Model: y = 0.2107x1 + 0.09676x2 + 3.912e-10
Epoch 40 Loss=0.2778 Model: y = 0.2655x1 + 0.1167x2 + -1.863e-11
Epoch 50 Loss=0.2512 Model: y = 0.3142x1 + 0.1321x2 + 1.77e-10
Epoch 60 Loss=0.2306 Model: y = 0.3576x1 + 0.1436x2 + -4.47e-10
Epoch 70 Loss=0.2145 Model: y = 0.3965x1 + 0.1519x2 + -8.103e-10
Epoch 80 Loss=0.2018 Model: y = 0.4313x1 + 0.1574x2 + -6.985e-10
Epoch 90 Loss=0.1917 Model: y = 0.4626x1 + 0.1607x2 + -4.936e-10
Epoch 100 Loss=0.1835 Model: y = 0.4909x1 + 0.1621x2 + -6.147e-10
Epoch 110 Loss=0.1769 Model: y = 0.5165x1 + 0.162x2 + -7.87e-10
Epoch 120 Loss=0.1714 Model: y = 0.5397x1 + 0.1606x2 + -5.821e-10
Epoch 130 Loss=0.1668 Model: y = 0.5609x1 + 0.1581x2 + -9.08e-10
Epoch 140 Loss=0.1629 Model: y = 0.5802x1 + 0.1549x2 + -9.965e-10
Epoch 150 Loss=0.1596 Model: y = 0.5979x1 + 0.1509x2 + -9.756e-10
Epoch 160 Loss=0.1567 Model: y = 0.6142x1 + 0.1465x2 + -4.144e-10
Epoch 170 Loss=0.1542 Model: y = 0.6292x1 + 0.1416x2 + -1.001e-10
Epoch 180 Loss=0.152 Model: y = 0.643x1 + 0.1364x2 + -3.236e-10
Epoch 190 Loss=0.15 Model: y = 0.6559x1 + 0.131x2 + -6.286e-11
Epoch 200 Loss=0.1483 Model: y = 0.6678x1 + 0.1255x2 + 2.119e-10
Epoch 210 Loss=0.1467 Model: y = 0.6789x1 + 0.1199x2 + -1.956e-10
Epoch 220 Loss=0.1453 Model: y = 0.6892x1 + 0.1142x2 + -1.758e-10
Epoch 230 Loss=0.144 Model: y = 0.6989x1 + 0.1085x2 + -4.307e-11
Epoch 240 Loss=0.1429 Model: y = 0.708x1 + 0.1029x2 + 3.376e-10
Epoch 250 Loss=0.1419 Model: y = 0.7165x1 + 0.09736x2 + 2.841e-10
Epoch 260 Loss=0.1409 Model: y = 0.7245x1 + 0.09189x2 + 3.295e-10
Epoch 270 Loss=0.14 Model: y = 0.732x1 + 0.08653x2 + -8.033e-11
Epoch 280 Loss=0.1393 Model: y = 0.7391x1 + 0.08128x2 + 1.141e-10
Epoch 290 Loss=0.1385 Model: y = 0.7458x1 + 0.07616x2 + 1.321e-10
Epoch 300 Loss=0.1379 Model: y = 0.7522x1 + 0.07118x2 + 5.087e-10
Epoch 310 Loss=0.1373 Model: y = 0.7582x1 + 0.06634x2 + 7.398e-10
Epoch 320 Loss=0.1367 Model: y = 0.7639x1 + 0.06165x2 + 6.845e-10
Epoch 330 Loss=0.1362 Model: y = 0.7693x1 + 0.0571x2 + 8.423e-10
Epoch 340 Loss=0.1358 Model: y = 0.7744x1 + 0.0527x2 + 9.252e-10
Epoch 350 Loss=0.1353 Model: y = 0.7793x1 + 0.04845x2 + 1.104e-09
Epoch 360 Loss=0.135 Model: y = 0.784x1 + 0.04435x2 + 1.145e-09
Epoch 370 Loss=0.1346 Model: y = 0.7884x1 + 0.0404x2 + 1.631e-09
Epoch 380 Loss=0.1343 Model: y = 0.7926x1 + 0.03658x2 + 1.446e-09
Epoch 390 Loss=0.134 Model: y = 0.7966x1 + 0.03291x2 + 1.429e-09
Epoch 400 Loss=0.1337 Model: y = 0.8004x1 + 0.02938x2 + 1.694e-09
Epoch 410 Loss=0.1335 Model: y = 0.8041x1 + 0.02598x2 + 1.697e-09
Epoch 420 Loss=0.1332 Model: y = 0.8076x1 + 0.02271x2 + 2.125e-09
Epoch 430 Loss=0.133 Model: y = 0.8109x1 + 0.01957x2 + 2.292e-09
Epoch 440 Loss=0.1328 Model: y = 0.8141x1 + 0.01655x2 + 2.913e-09
Epoch 450 Loss=0.1327 Model: y = 0.8171x1 + 0.01366x2 + 3.412e-09
Epoch 460 Loss=0.1325 Model: y = 0.82x1 + 0.01087x2 + 3.749e-09
Epoch 470 Loss=0.1323 Model: y = 0.8228x1 + 0.008204x2 + 3.499e-09
Epoch 480 Loss=0.1322 Model: y = 0.8254x1 + 0.005641x2 + 3.663e-09
Epoch 490 Loss=0.1321 Model: y = 0.828x1 + 0.003183x2 + 4.2e-09
Epoch 500 Loss=0.132 Model: y = 0.8304x1 + 0.0008239x2 + 4.138e-09