刚开始学tensorflow。。现在有将近1万条数据，每条数据有两个属性。想用一个属性来预测另一个属性。。看网上的代码写了个线性回归。但是训练起来好慢。。50个batch要训练10分钟多一点。。我的batch大小是1（主要是不知道batch有啥用），显卡是1060的。 想问问大佬们这正常吗

from __future__ import print_function, division
import tensorflow as tf
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn
# 读入数据
train = pd.read_csv("1.csv")
# 选取房屋面积小于１２０００的数据
train = train.dropna()
train_X = train['NA_Sales'].values.reshape(-1, 1)
train_Y = train['Global_Sales'].values.reshape(-1, 1)
n_samples = train_X.shape[0]
# 学习率
learning_rate = 2
# 迭代次数
training_epochs = 1000
# 每多少次输出一次迭代结果
display_step = 50
# 这个X和Y和上面的train_X,train_Y是不一样的，这里只是个占位符，
# 训练开始的时候需要“喂”(feed)数据给它
X = tf.placeholder(tf.float32)
Y = tf.placeholder(tf.float32)
# 定义模型参数
W = tf.Variable(np.random.randn(), name="weight", dtype=tf.float32)
b = tf.Variable(np.random.randn(), name="bias", dtype=tf.float32)
# 定义模型
pred = tf.add(tf.multiply(W, X), b)
# 定义损失函数
cost = tf.reduce_sum(tf.pow(pred-Y, 2)) / (2 * n_samples)
# 使用Adam算法，至于为什么不使用一般的梯度下降算法，一会说
optimizer = tf.train.AdamOptimizer(learning_rate).minimize(cost)
# 初始化所有变量
init = tf.initialize_all_variables()
# 训练开始
with tf.Session() as sess:
sess.run(init)
for epoch in range(training_epochs):
for (x, y) in zip(train_X, train_Y):
sess.run(optimizer, feed_dict={X: x, Y: y})
if (epoch + 1) % display_step == 0:
c = sess.run(cost, feed_dict={X: train_X, Y: train_Y})
print("Epoch:", '%04d' % (epoch + 1), "cost=", "{:.3f}".format(c), "W=", sess.run(W), "b=", sess.run(b))
print("Optimization Finished!")
training_cost = sess.run(cost, feed_dict={X: train_X, Y: train_Y})
print("Training cost=", training_cost, "W=", sess.run(W), "b=", sess.run(b), '\n')
# 画图
plt.plot(train_X, train_Y, 'ro', label="Original data")
plt.plot(train_X, sess.run(W) * train_X + sess.run(b), label="Fitted line")
plt.legend()
plt.show()

楼主，我也在学习线性回归这部分的知识。可以请教你一个问题吗？
看不懂这里怎么就可以预测了呢
linear_regressor = tf.estimator.LinearRegressor(
feature_columns=feature_columns,
optimizer=my_optimizer
)
#训练模型
_ = linear_regressor.train(
input_fn = lambda:my_input_fn(my_feature, targets),
steps=100
)
#预测模型正确性
prediction_input_fn =lambda: my_input_fn(my_feature, targets, num_epochs=1, shuffle=False)
# Call predict() on the linear_regressor to make predictions.
predictions = linear_regressor.predict(input_fn=prediction_input_fn)
# Format predictions as a NumPy array, so we can calculate error metrics.
predictions = np.array([item['predictions'][0] for item in predictions])