Long Short-Term Memory network is a type of Recurrent Neural Network


In [2]:
import numpy
import matplotlib.pyplot as plt
import pandas
import math
from keras.models import Sequential
from keras.layers import Dense
from keras.layers import LSTM
from sklearn.preprocessing import MinMaxScaler
from sklearn.metrics import mean_squared_error


Using Theano backend.
WARNING (theano.configdefaults): g++ not detected ! Theano will be unable to execute optimized C-implementations (for both CPU and GPU) and will default to Python implementations. Performance will be severely degraded. To remove this warning, set Theano flags cxx to an empty string.

In [3]:
# fix random seed for reproducibility
numpy.random.seed(7)

In [4]:
# load the dataset
dataframe = pandas.read_csv('netreachvolumes.csv', usecols=[1], engine='python')
dataset = dataframe.values
dataset = dataset.astype('float32')

In [5]:
dataframe.describe()


Out[5]:
Net Reach
count 1.710000e+02
mean 9.597096e+06
std 6.485620e+06
min 1.035764e+06
25% 5.870565e+06
50% 8.173524e+06
75% 1.103199e+07
max 4.972979e+07

In [6]:
# normalize the dataset
scaler = MinMaxScaler(feature_range=(0, 1))
dataset = scaler.fit_transform(dataset)

In [7]:
# split into train and test sets - train set ends on May 12
train_size = 61
test_size = len(dataset) - train_size
train, test = dataset[0:train_size,:], dataset[train_size-2:len(dataset),:]
print(len(train), len(test))


(61, 112)

In [8]:
def create_dataset(dataset, look_back=1):
    dataX, dataY = [], []
    for i in range(len(dataset)-look_back-1):
        a = dataset[i:(i+look_back), 0]
        dataX.append(a)
        dataY.append(dataset[i + look_back, 0])
    return numpy.array(dataX), numpy.array(dataY)

In [9]:
# reshape into X=t and Y=t+1
look_back = 1
trainX, trainY = create_dataset(train, look_back)
testX, testY = create_dataset(test, look_back)

In [10]:
# reshape input to be [samples, time steps, features]
trainX = numpy.reshape(trainX, (trainX.shape[0], 1, trainX.shape[1]))
testX = numpy.reshape(testX, (testX.shape[0], 1, testX.shape[1]))

In [11]:
# create and fit the LSTM network
model = Sequential()
model.add(LSTM(4, input_dim=look_back))
model.add(Dense(1))
model.compile(loss='mean_squared_error', optimizer='adam')
model.fit(trainX, trainY, nb_epoch=10, batch_size=1, verbose=2)


Epoch 1/10
1s - loss: 0.0594
Epoch 2/10
1s - loss: 0.0346
Epoch 3/10
1s - loss: 0.0231
Epoch 4/10
1s - loss: 0.0189
Epoch 5/10
1s - loss: 0.0178
Epoch 6/10
1s - loss: 0.0177
Epoch 7/10
1s - loss: 0.0176
Epoch 8/10
1s - loss: 0.0176
Epoch 9/10
1s - loss: 0.0175
Epoch 10/10
1s - loss: 0.0175
Out[11]:
<keras.callbacks.History at 0x11c4b630>

In [12]:
# make predictions
trainPredict = model.predict(trainX)
testPredict = model.predict(testX)
# invert predictions
trainPredict = scaler.inverse_transform(trainPredict)
trainY = scaler.inverse_transform([trainY])
testPredict = scaler.inverse_transform(testPredict)
testY = scaler.inverse_transform([testY])
# calculate root mean squared error
trainScore = math.sqrt(mean_squared_error(trainY[0], trainPredict[:,0]))
print('Train Score: %.2f RMSE' % (trainScore))
testScore = math.sqrt(mean_squared_error(testY[0], testPredict[:,0]))
print('Test Score: %.2f RMSE' % (testScore))


Train Score: 6420830.35 RMSE
Test Score: 6777212.30 RMSE

In [13]:
# shift train predictions for plotting
trainPredictPlot = numpy.empty_like(dataset)
trainPredictPlot[:, :] = numpy.nan
trainPredictPlot[look_back:len(trainPredict)+look_back, :] = trainPredict
# shift test predictions for plotting
testPredictPlot = numpy.empty_like(dataset)
testPredictPlot[:, :] = numpy.nan
testPredictPlot[len(trainPredict)+(look_back*2)+1-2:len(dataset)-1, :] = testPredict
# plot baseline and predictions
plt.plot(scaler.inverse_transform(dataset))
plt.plot(trainPredictPlot)
plt.plot(testPredictPlot)
#plt.show()


Out[13]:
[<matplotlib.lines.Line2D at 0x118fe2b0>]

In [14]:
trainPredictPlot[59]


Out[14]:
array([ 14032136.], dtype=float32)

In [15]:
testPredictPlot[59]


Out[15]:
array([ nan], dtype=float32)

In [16]:
import numpy as np

In [17]:
np.count_nonzero(~np.isnan(testPredictPlot))


Out[17]:
110

In [18]:
testPredictPlot[60]


Out[18]:
array([ 14966257.], dtype=float32)

In [19]:
trainPredictPlot[60:] = testPredictPlot[60:]

In [20]:
trainPredictPlot


Out[20]:
array([[       nan],
       [ 12249573.],
       [ 12651656.],
       [ 12185246.],
       [ 12483526.],
       [ 13020171.],
       [ 12860371.],
       [ 13791076.],
       [ 13060127.],
       [ 12671042.],
       [ 12705147.],
       [ 12709331.],
       [ 12601954.],
       [ 12860981.],
       [ 12420385.],
       [ 12212492.],
       [ 12299818.],
       [ 12219980.],
       [ 12227456.],
       [ 12235809.],
       [ 12417684.],
       [ 12356925.],
       [ 12176556.],
       [ 12219850.],
       [ 12256937.],
       [ 12478849.],
       [ 12461597.],
       [ 12371209.],
       [ 13267556.],
       [ 13180627.],
       [ 13446882.],
       [ 13016415.],
       [ 13645996.],
       [ 13109773.],
       [ 13068709.],
       [ 12899793.],
       [ 12449484.],
       [ 12488706.],
       [ 13134220.],
       [ 12344841.],
       [ 12942468.],
       [ 14866024.],
       [ 18031936.],
       [ 12518429.],
       [ 12778576.],
       [ 13126175.],
       [ 12908876.],
       [ 16821202.],
       [ 12735309.],
       [ 12629929.],
       [ 13018110.],
       [ 12591983.],
       [ 13815635.],
       [ 12837222.],
       [ 13815490.],
       [ 13232275.],
       [ 13016568.],
       [ 12746867.],
       [ 13757101.],
       [ 14032136.],
       [ 14966257.],
       [ 14762188.],
       [ 19076098.],
       [ 15466003.],
       [ 13713550.],
       [ 12368828.],
       [ 12325559.],
       [ 12364280.],
       [ 12416262.],
       [ 13040218.],
       [ 12667853.],
       [ 12420176.],
       [ 12095921.],
       [ 12442565.],
       [ 11843852.],
       [ 12273151.],
       [ 12392452.],
       [ 12241090.],
       [ 12681466.],
       [ 14478686.],
       [ 12929304.],
       [ 12146098.],
       [ 12156492.],
       [ 12542082.],
       [ 12175720.],
       [ 12216788.],
       [ 12700426.],
       [ 12733682.],
       [ 12432803.],
       [ 12660229.],
       [ 12274866.],
       [ 12388720.],
       [ 12174144.],
       [ 12550003.],
       [ 11941202.],
       [ 12170595.],
       [ 12269610.],
       [ 12448387.],
       [ 12630419.],
       [ 12429844.],
       [ 13012295.],
       [ 12205902.],
       [ 12443660.],
       [ 12023237.],
       [ 12087981.],
       [ 11925878.],
       [ 12835016.],
       [ 12236864.],
       [ 12078882.],
       [ 12647809.],
       [ 12073921.],
       [ 12040182.],
       [ 11949301.],
       [ 12197095.],
       [ 11859483.],
       [ 12170792.],
       [ 11842327.],
       [ 11981870.],
       [ 12805680.],
       [ 12022976.],
       [ 12633114.],
       [ 11789864.],
       [ 11741246.],
       [ 12141356.],
       [ 11966613.],
       [ 11677493.],
       [ 11790964.],
       [ 12021489.],
       [ 12117640.],
       [ 11852034.],
       [ 12045219.],
       [ 11930291.],
       [ 12747446.],
       [ 12062887.],
       [ 11916695.],
       [ 11888655.],
       [ 11852149.],
       [ 11898581.],
       [ 11899027.],
       [ 12282441.],
       [ 12121412.],
       [ 11768130.],
       [ 11983410.],
       [ 11921796.],
       [ 12116822.],
       [ 11913852.],
       [ 12639032.],
       [ 11980227.],
       [ 11864072.],
       [ 12074951.],
       [ 11841354.],
       [ 11751408.],
       [ 12343355.],
       [ 12356552.],
       [ 12283379.],
       [ 12553230.],
       [ 12020538.],
       [ 12276724.],
       [ 11716824.],
       [ 12336205.],
       [ 12561706.],
       [ 11850090.],
       [ 11897272.],
       [ 11745280.],
       [ 11820488.],
       [ 11840401.],
       [ 11832390.],
       [ 12169358.],
       [ 11822655.],
       [ 11905107.],
       [       nan]], dtype=float32)

In [21]:
dataset = scaler.inverse_transform(dataset)

In [22]:
trainPredictPlot[0] = dataset[0]

In [23]:
line_up, = plt.plot(trainPredictPlot, label='predicted')
line_down, = plt.plot(dataset, label='observed')
plt.legend(labels=[line_up, line_down])
plt.show()

In [24]:
differences = trainPredictPlot -dataset

In [25]:
import numpy
differences = numpy.asarray(differences)

In [26]:
import pandas as pd
df = pd.read_csv('netreachvolumes.csv')
df.columns = [['date','volume']]
df['predicted'] = trainPredictPlot
df['difference'] = differences

In [27]:
df.to_csv("netreachdifferences.csv")

In [ ]: