Question

我正在尝试使用this教程为多个变量实现线性回归模型。我尝试使用sklearn的train_test_split()方法将其替换为自己的数据集。

import numpy as np
import tensorflow as tf
import pandas as pd

df = pd.read_csv('airfoil_self_noise.csv',sep=',')
from sklearn.model_selection import train_test_split
X_true, X_test, y_true, y_test = train_test_split(df.iloc[:,:-1].values,df.iloc[:,-1].values,test_size = 0.2, random_state=0)

n_features = np.shape(X_true)[1]
m_examples = np.shape(X_true)[0]

# Placeholder that is fed input data.
X_in = tf.placeholder(tf.float32, [None, n_features], "X_in")

# The model: we assume y = X_in * w + b
w = tf.Variable(tf.random_normal((n_features, 1)), name="w")
b = tf.Variable(tf.constant(0.1, shape=[]), name="b")
h = tf.add(tf.matmul(X_in, w), b, name="h")

# Placeholder that is fed observed results.
y_in = tf.placeholder(tf.float32,[1,None], "y_in")

# The loss function: we are minimizing square root of mean 
loss_op = tf.reduce_mean(tf.square(tf.subtract(y_in, h)), name="loss")
train_op = tf.train.GradientDescentOptimizer(0.3).minimize(loss_op)

with tf.Session() as sess:
    sess.run(tf.global_variables_initializer())
    for step in range(1000):
        sess.run(train_op, feed_dict={
          X_in: X_true, 
          y_in: y_true
        })
    w_computed = sess.run(w)
    b_computed = sess.run(b)


print ("w computed [%s]" % ', '.join(['%.5f' % x for x in w_computed.flatten()]))
print ("w actual   [%s]" % ', '.join(['%.5f' % x for x in w_true.flatten()]))
print ("b computed %.3f" % b_computed)
print ("b actual  %.3f" % b_true[0])

我似乎遇到的问题是送入y_in的numpy数组的形状。

Traceback (most recent call last):
  File "Airfoil_Test_TF.py", line 32, in <module>
    y_in: y_true
  File ".../anaconda3/envs/py36/lib/python3.6/site-packages/tensorflow/python/client/session.py", line 929, in run
    run_metadata_ptr)
  File ".../anaconda3/envs/py36/lib/python3.6/site-packages/tensorflow/python/client/session.py", line 1128, in _run
    str(subfeed_t.get_shape())))
ValueError: Cannot feed value of shape (1202,) for Tensor 'y_in:0', which has shape '(1, ?)'

我试图修改y_in的占位符的尺寸，但是它没有任何作用。本教程最初使用尺寸[None,1]定义了占位符，而不是采用这种方式，但是我找不到找到将y_true转换为形状（，1202）的方法，因为一个维数组无法被换成numpy。

有什么建议吗？

谢谢！

Answer 1

您可以致电

y_true = y_true.reshape((1,-1))

这应该可以解决问题。

简短说明：

在numpy中，形状(10,)只是一维向量。
形状(10, 1)代表一个明确的行向量。
形状(1, 10)代表列向量。

Numpy数组的feed_dict形状出现问题

1 个答案: