用神经网络预测股票价格，tensorflow代码实现

1. # layer1：激励函数+乘加运算

2. import tensorflow as tf

3. import numpy as np

4. import matplotlib.pyplot as plt

5. date = np.linspace(1,15,15)

6. endPrice = np.array([2511.90,2538.26,2510.68,2591.66,2732.98,2701.69,2701.29,2678.67,2726.50,2681.50,2739.17,2715.07,2823.58,2864.90,2919.08]

7. )

8. beginPrice = np.array([2438.71,2500.88,2534.95,2512.52,2594.04,2743.26,2697.47,2695.24,2678.23,2722.13,2674.93,2744.13,2717.46,2832.73,2877.40])

9. print(date)

10. plt.figure()

11. for i in range(0,15):

12. # 1 柱状图

13. dateOne = np.zeros([2])

14. dateOne[0] = i;

15. dateOne[1] = i;

16. priceOne = np.zeros([2])

17. priceOne[0] = beginPrice[i]

18. priceOne[1] = endPrice[i]

19. if endPrice[i]>beginPrice[i]:

20. plt.plot(dateOne,priceOne,'r',lw=8)

21. else:

22. plt.plot(dateOne,priceOne,'g',lw=8)

23. #plt.show()

24. # A(15x1)*w1(1x10)+b1(1*10) = B(15x10)

25. # B(15x10)*w2(10x1)+b2(15x1) = C(15x1)

26. # 1 A B C

27. dateNormal = np.zeros([15,1])

28. priceNormal = np.zeros([15,1])

29. #归一化

30. for i in range(0,15):

31. dateNormal[i,0] = i/14.0;

32. priceNormal[i,0] = endPrice[i]/3000.0;

33. x = tf.placeholder(tf.float32,[None,1])

34. y = tf.placeholder(tf.float32,[None,1])

35. # B

36. w1 = tf.Variable(tf.random_uniform([1,10],0,1))

37. b1 = tf.Variable(tf.zeros([1,10]))

38. wb1 = tf.matmul(x,w1)+b1

39. layer1 = tf.nn.relu(wb1) # 激励函数

40. # C

41. w2 = tf.Variable(tf.random_uniform([10,1],0,1))

42. b2 = tf.Variable(tf.zeros([15,1]))

43. wb2 = tf.matmul(layer1,w2)+b2

44. layer2 = tf.nn.relu(wb2)

45. loss = tf.reduce_mean(tf.square(y-layer2))#y 真实 layer2 计算

46. #梯度下级法

47. train_step = tf.train.GradientDescentOptimizer(0.1).minimize(loss)

48. with tf.Session() as sess:

49. sess.run(tf.global_variables_initializer())

50. for i in range(0,10000):

51. sess.run(train_step,feed_dict={x:dateNormal,y:priceNormal})

52. #预测下

53. # w1w2 b1b2 A + wb -->layer2

54. pred = sess.run(layer2,feed_dict={x:dateNormal})

55. predPrice = np.zeros([15,1])

56. for i in range(0,15):

57. predPrice[i,0]=(pred*3000)[i,0]

58. plt.plot(date,predPrice,'b',lw=1)

59. plt.show()