【Tensorflow】Python实现神经网络回归

柔情只为你懂 2022-05-31 14:06 299阅读 0赞

### 环境 ###

macOS，python3.6，tensorflow1.1.0

### 回归问题 ###

<table> 
 <thead> 
  <tr> 
   <th align="left">年份</th> 
   <th align="right">事故起数</th> 
   <th align="right">死亡人数</th> 
   <th align="right">受伤人数</th> 
   <th align="right">直接财产损失（万元）</th> 
  </tr> 
 </thead> 
 <tbody> 
  <tr> 
   <td align="left">2003</td> 
   <td align="right">80</td> 
   <td align="right">3</td> 
   <td align="right">50</td> 
   <td align="right">11</td> 
  </tr> 
  <tr> 
   <td align="left">2004</td> 
   <td align="right">90</td> 
   <td align="right">8</td> 
   <td align="right">70</td> 
   <td align="right">12.5</td> 
  </tr> 
  <tr> 
   <td align="left">2005</td> 
   <td align="right">180</td> 
   <td align="right">20</td> 
   <td align="right">120</td> 
   <td align="right">20</td> 
  </tr> 
  <tr> 
   <td align="left">2006</td> 
   <td align="right">140</td> 
   <td align="right">16</td> 
   <td align="right">90</td> 
   <td align="right">18</td> 
  </tr> 
  <tr> 
   <td align="left">2007</td> 
   <td align="right">120</td> 
   <td align="right">5</td> 
   <td align="right">80</td> 
   <td align="right">15</td> 
  </tr> 
 </tbody> 
</table>

说明：利用前四年的数据建立回归模型，并对第五年进行预测。

### 实现代码 ###

import numpy as np
    import tensorflow as tf
    
    x = [[80,3,50],[90,8,70],[180,20,120],[140,16,90]]
    y = [[11],[12.5],[20],[18]]
    # y = [11,12.5,20,18]
    x_pred = [[120,5,85]]
    
    tf_x = tf.placeholder(tf.float32, [None,3])     # input x
    tf_y = tf.placeholder(tf.float32, [None,1])     # input y
    
    # neural network layers
    l1 = tf.layers.dense(tf_x, 20, tf.nn.relu)          # hidden layer
    output = tf.layers.dense(l1, 1)                     # output layer
    
    loss = tf.losses.mean_squared_error(tf_y, output)   # compute cost
    optimizer = tf.train.GradientDescentOptimizer(learning_rate=0.1)
    train_op = optimizer.minimize(loss)
    
    sess = tf.Session()                                 # control training and others
    sess.run(tf.global_variables_initializer())         # initialize var in graph
    
    
    for step in range(150):
        # train and net output
        _, l, pred = sess.run([train_op, loss, output], {tf_x: x, tf_y: y})
        if step % 10 == 0:
            print('loss is: ' + str(l))
            # print('prediction is:' + str(pred))
    
    output_pred = sess.run(output,{tf_x:x_pred})
    print('input is:' + str(x_pred[0][:]))
    print('output is:' + str(output_pred[0][0]))

### 运行结果 ###

![这里写图片描述][70]

[70]: /images/20220531/c730623206b547f590d2780bb63c0380.png