Added LSTM to data prediction

klab-12-3-rnn_prediction.py

@@ -1,7 +1,7 @@
 # http://machinelearningmastery.com/time-series-prediction-lstm-recurrent-neural-networks-python-keras/
 import numpy as np
 from keras.models import Sequential
-from keras.layers import Dense, LSTM
+from keras.layers import Dense, LSTM, Activation
 from sklearn.preprocessing import MinMaxScaler
 import os
 
@@ -17,7 +17,7 @@
 
 import matplotlib.pyplot as plt
 # Open,High,Low,Close,Volume
-xy = np.loadtxt('data-02-data-02-stock_daily.csv', delimiter=',')
+xy = np.loadtxt('data-02-stock_daily.csv', delimiter=',')
 xy = xy[::-1]  # reverse order (chronically ordered)
 
 # very important. It does not work without it.
@@ -45,8 +45,9 @@
     dataY[train_size:len(dataY)])
 
 model = Sequential()
-model.add(LSTM(5, input_shape=(timesteps, data_dim), return_sequences=False))
-model.add(Dense(1))
+model.add(LSTM(1, input_shape=(seq_length, data_dim), return_sequences=False))
+#model.add(Dense(1))
+model.add(Activation("linear"))
 model.compile(loss='mean_squared_error', optimizer='adam')
 
 model.summary()
@@ -64,7 +65,7 @@
 #testPredict = scaler.transform(testPredict)
 #testY = scaler.transform(testY)
 
-print(testPredict)
-# plt.plot(testY)
-# plt.plot(testPredict)
-# plt.show()
+#print(testPredict)
+plt.plot(testY)
+plt.plot(testPredict)
+plt.show()

klab-12-4-rnn_2_prediction.py

@@ -18,7 +18,7 @@
 
 import matplotlib.pyplot as plt
 # Open,High,Low,Close,Volume
-xy = np.loadtxt('data-02-data-02-stock_daily.csv', delimiter=',')
+xy = np.loadtxt('data-02-stock_daily.csv', delimiter=',')
 xy = xy[::-1]  # reverse order (chronically ordered)
 
 # very important. It does not work without it.
@@ -70,6 +70,6 @@
 #testY = scaler.transform(testY)
 
 print(testPredict)
-# plt.plot(testY)
-# plt.plot(testPredict)
-# plt.show()
+plt.plot(testY)
+plt.plot(testPredict)
+plt.show()

lab-12-4-rnn_stock_prediction.py

@@ -0,0 +1,72 @@
+import tensorflow as tf
+import numpy as np
+import matplotlib.pyplot as plt
+from sklearn.preprocessing import MinMaxScaler
+
+tf.set_random_seed(777)  # reproducibility
+
+timesteps = seq_length = 7
+data_dim = 5
+output_dim = 1
+
+# Open,High,Low,Close,Volume
+xy = np.loadtxt('data-02-stock_daily.csv', delimiter=',')
+xy = xy[::-1]  # reverse order (chronically ordered)
+
+# very important. It does not work without it.
+scaler = MinMaxScaler(feature_range=(0, 1))
+xy = scaler.fit_transform(xy)
+
+x = xy
+y = xy[:, [-1]]  # Close as label
+
+dataX = []
+dataY = []
+for i in range(0, len(y) - seq_length):
+    _x = x[i:i + seq_length]
+    _y = y[i + 1]  # Next close price
+    print(_x, "->", _y)
+    dataX.append(_x)
+    dataY.append(_y)
+
+# split to train and testing
+train_size = int(len(dataY) * 0.7)
+test_size = len(dataY) - train_size
+trainX, testX = np.array(dataX[0:train_size]), np.array(
+    dataX[train_size:len(dataX)])
+trainY, testY = np.array(dataY[0:train_size]), np.array(
+    dataY[train_size:len(dataY)])
+
+# input place holders
+X = tf.placeholder(tf.float32, [None, seq_length, data_dim])
+Y = tf.placeholder(tf.float32, [None, 1])
+
+cell = tf.contrib.rnn.BasicLSTMCell(num_units=output_dim, state_is_tuple=True)
+outputs, _states = tf.nn.dynamic_rnn(cell, X, dtype=tf.float32)
+Y_pred = outputs[:, -1] # We use the last cell's output
+
+print(outputs[:,-1])
+
+# loss
+loss = tf.reduce_sum(tf.square(Y_pred - Y))  # sum of the squares
+# optimizer
+optimizer = tf.train.GradientDescentOptimizer(0.01)
+train = optimizer.minimize(loss)
+
+# RMSE
+targets = tf.placeholder(tf.float32, [None, 1])
+predictions = tf.placeholder(tf.float32, [None, 1])
+rmse = tf.sqrt(tf.reduce_mean(tf.square(targets - predictions)))
+
+sess = tf.Session()
+sess.run(tf.global_variables_initializer())
+
+for i in range(1000):
+    _, l = sess.run([train, loss], feed_dict={X: trainX, Y:trainY})
+    print(i, l)
+
+testPredict = sess.run(Y_pred, feed_dict={X: testX})
+print("RMSE", sess.run(rmse, feed_dict={targets: testY, predictions:testPredict}))
+plt.plot(testY)
+plt.plot(testPredict)
+plt.show()

lab-12-5-rnn_softmax_stock_prediction.py

@@ -0,0 +1,76 @@
+import tensorflow as tf
+import numpy as np
+import matplotlib.pyplot as plt
+from sklearn.preprocessing import MinMaxScaler
+
+tf.set_random_seed(777)  # reproducibility
+
+timesteps = seq_length = 7
+data_dim = 5
+output_dim = 3
+
+# Open,High,Low,Close,Volume
+xy = np.loadtxt('data-02-stock_daily.csv', delimiter=',')
+xy = xy[::-1]  # reverse order (chronically ordered)
+
+# very important. It does not work without it.
+scaler = MinMaxScaler(feature_range=(0, 1))
+xy = scaler.fit_transform(xy)
+
+x = xy
+y = xy[:, [-1]]  # Close as label
+
+dataX = []
+dataY = []
+for i in range(0, len(y) - seq_length):
+    _x = x[i:i + seq_length]
+    _y = y[i + 1]  # Next close price as target
+    print(_x, "->", _y)
+    dataX.append(_x)
+    dataY.append(_y)
+
+# split to train and testing
+train_size = int(len(dataY) * 0.7)
+test_size = len(dataY) - train_size
+trainX, testX = np.array(dataX[0:train_size]), np.array(
+    dataX[train_size:len(dataX)])
+trainY, testY = np.array(dataY[0:train_size]), np.array(
+    dataY[train_size:len(dataY)])
+
+# input place holders
+X = tf.placeholder(tf.float32, [None, seq_length, data_dim])
+Y = tf.placeholder(tf.float32, [None, 1])
+
+cell = tf.contrib.rnn.BasicLSTMCell(num_units=output_dim, state_is_tuple=True)
+cell = tf.contrib.rnn.MultiRNNCell([cell] * 2, state_is_tuple=True)
+outputs, _states = tf.nn.dynamic_rnn(cell, X, dtype=tf.float32)
+last_output = outputs[:, -1] # We use the last cell's output
+
+# Softmax layer (rnn_hidden_size -> num_classes)
+softmax_w = tf.get_variable("softmax_w", [output_dim, 1])
+softmax_b = tf.get_variable("softmax_b", [1])
+Y_pred = tf.matmul(last_output,  softmax_w) + softmax_b
+
+# loss
+loss = tf.reduce_sum(tf.square(Y_pred - Y))  # sum of the squares
+# optimizer
+optimizer = tf.train.GradientDescentOptimizer(0.001)
+train = optimizer.minimize(loss)
+
+# RMSE
+targets = tf.placeholder(tf.float32, [None, 1])
+predictions = tf.placeholder(tf.float32, [None, 1])
+rmse = tf.sqrt(tf.reduce_mean(tf.square(targets - predictions)))
+
+sess = tf.Session()
+sess.run(tf.global_variables_initializer())
+
+for i in range(1000):
+    _, l = sess.run([train, loss], feed_dict={X: trainX, Y:trainY})
+    print(i, l)
+
+testPredict = sess.run(Y_pred, feed_dict={X: testX})
+print("RMSE", sess.run(rmse, feed_dict={targets: testY, predictions:testPredict}))
+plt.plot(testY)
+plt.plot(testPredict)
+plt.show()

tests/test_square.py

@@ -24,5 +24,11 @@ def testBroadcast(self):
             cost = tf.reduce_mean(tf.square(hypothesis - y))
             self.assertAllEqual(cost.eval(), 0)
 
+    def testNormalize(self):
+        with self.test_session():
+            values = np.array([[10, 20], [1000, -100]], dtype=np.float32)
+            norm_values = tf.nn.l2_normalize(values, dim=1)
+            print(norm_values.eval())
+
 if __name__ == '__main__':
     tf.test.main()