LSTM
----

.. code:: ipython3

    import pandas as pd
    import numpy as np
    import matplotlib.pyplot as plt
    from keras.models import Sequential
    from keras.layers import Dense, SimpleRNN, LSTM, Dropout
    from sklearn.preprocessing import MinMaxScaler
    from sklearn.metrics import mean_squared_error, r2_score
    from sklearn import linear_model

.. code:: ipython3

    data = pd.read_csv("TRM.csv", sep = ";", decimal=",")
    dataNorm = data.drop(['Fecha'], 1, inplace=True)
    data.plot()
    plt.show()



.. image:: output_2_0.png


Conjunto de Train y Test:
~~~~~~~~~~~~~~~~~~~~~~~~~

200 precios como test.

.. code:: ipython3

    train = data.values[:len(data) - 200]
    test = data.values[len(data) - 200:]

Estandarización:
~~~~~~~~~~~~~~~~

.. code:: ipython3

    sc = MinMaxScaler()

.. code:: ipython3

    train = sc.fit_transform(train)
    test = sc.fit_transform(test)

Creación datos de entrada a la red:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Los datos de entrada a la red tienen que tener la forma de 3D tensor:

.. math::  [samples, timesteps, features] 

La siguiente función crea los times steps.

.. code:: ipython3

    def batch(data, timeSteps):
     X, Y =[], []
     for i in range(len(data)-timeSteps):
      j=i+timeSteps  
      X.append(data[i:j,0])
      Y.append(data[j,0])
     return np.array(X), np.array(Y)

Time steps de 5 y features de 1, es decir, se utilizarán 5 precios
históricos para predecir 1 precio.

.. code:: ipython3

    X_train, y_train = batch(train, 5)
    X_test, y_test = batch(test, 5)

Transformación de los datos en 3D tensor:

.. code:: ipython3

    X_train = np.reshape(X_train, (X_train.shape[0], X_train.shape[1], 1))
    X_test = np.reshape(X_test, (X_test.shape[0], X_test.shape[1], 1))

RNN:
----

.. code:: ipython3

    model = Sequential()
    model.add(SimpleRNN(units = 10, activation = "relu", input_shape=(5,1)))
    model.add(Dropout(0.2))
    model.add(Dense(1))
    model.compile(optimizer = 'adam', loss = 'mean_squared_error')
    history = model.fit(X_train, y_train, validation_data=(X_test, y_test), batch_size = 20, epochs = 10, verbose=0)

Comparación Loss training vs. Loss testing:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. code:: ipython3

    epochs = range(1, 10 + 1)
    loss = history.history["loss"]
    valLoss = history.history["val_loss"]
    
    plt.plot(epochs, loss, "b", label = "Training Loss", color = "blue")
    plt.plot(epochs, valLoss, "b", label = "Testing Loss", color = "red")
    plt.title("Model Loss")
    plt.legend()
    plt.show()



.. image:: output_19_0.png


.. code:: ipython3

    y_test_pred_RNN = model.predict(X_test)
    y_test_pred_RNN = sc.inverse_transform(y_test_pred_RNN.reshape(-1,1))
    y_true = sc.inverse_transform(y_test.reshape(-1,1))

MSE
~~~

.. code:: ipython3

    mean_squared_error(y_true, y_test_pred_RNN)




.. parsed-literal::

    1117.1567997692296



.. code:: ipython3

    plt.figure(figsize=(12, 6), dpi=80, facecolor = 'w', edgecolor = 'k')
    
    plt.plot(y_true, color='black', label='Real')
    plt.plot(y_test_pred_RNN, color = 'green', label = 'RNN')
    
    plt.title('Real vs. RNN', fontsize = 12)
    plt.xlabel('Time (test)', fontsize=12)
    plt.ylabel('Price', fontsize = 12)
    plt.legend(loc = 'best')
    plt.show()



.. image:: output_23_0.png


Ajuste de regresión
~~~~~~~~~~~~~~~~~~~

.. code:: ipython3

    regressor = linear_model.LinearRegression()

.. code:: ipython3

    regressor.fit(y_true, y_test_pred_RNN)




.. parsed-literal::

    LinearRegression()



.. code:: ipython3

    print('R2: %.2f' % r2_score(y_true, y_test_pred_RNN))


.. parsed-literal::

    R2: 0.95
    

LSTM:
~~~~~

.. figure:: LSTM.jpg
   :alt: 1

   1

Una capa de 10 celdas LSTM.

Función de activación: ReLu.

Batch size: 20.

Epochs: 10.

Dropout: 20% para evitar overfitting.

.. code:: ipython3

    model = Sequential()
    model.add(LSTM(units = 10, activation = "relu", input_shape=(5,1)))
    model.add(Dropout(0.2))
    model.add(Dense(1))
    model.compile(optimizer = 'adam', loss = 'mean_squared_error')
    history = model.fit(X_train, y_train, validation_data=(X_test, y_test), batch_size = 20, epochs = 10, verbose=0)

Comparación Loss training vs. Loss testing:
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

.. code:: ipython3

    epochs = range(1, 10 + 1)
    loss = history.history["loss"]
    valLoss = history.history["val_loss"]
    
    plt.plot(epochs, loss, "b", label = "Training Loss", color = "blue")
    plt.plot(epochs, valLoss, "b", label = "Testing Loss", color = "red")
    plt.title("Model Loss")
    plt.legend()
    plt.show()



.. image:: output_33_0.png


.. code:: ipython3

    y_test_pred_LSTM = model.predict(X_test)
    y_test_pred_LSTM = sc.inverse_transform(y_test_pred_LSTM.reshape(-1,1))

MSE
~~~

.. code:: ipython3

    mean_squared_error(y_true, y_test_pred_LSTM)




.. parsed-literal::

    2618.8473693593573



.. code:: ipython3

    plt.figure(figsize=(12, 6), dpi=80, facecolor = 'w', edgecolor = 'k')
    
    plt.plot(y_true, color='black', label='Real')
    plt.plot(y_test_pred_LSTM, color = 'green', label = 'LSTM')
    
    plt.title('Real vs. LSTM', fontsize = 12)
    plt.xlabel('Time (test)', fontsize=12)
    plt.ylabel('Price', fontsize = 12)
    plt.legend(loc = 'best')
    plt.show()



.. image:: output_37_0.png


Ajuste de regresión
~~~~~~~~~~~~~~~~~~~

.. code:: ipython3

    regressor = linear_model.LinearRegression()

.. code:: ipython3

    regressor.fit(y_true, y_test_pred_LSTM)




.. parsed-literal::

    LinearRegression()



.. code:: ipython3

    print('R2: %.2f' % r2_score(y_true, y_test_pred_LSTM))


.. parsed-literal::

    R2: 0.88