dolar/dolar.py at master · acrosa/dolar · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
from matplotlib import pyplot
import numpy
import pandas as pd
import requests
import tensorflow as tf
from tensorflow import keras
from tensorflow.python.keras import Sequential
from tensorflow.python.keras.layers import LSTM, Dense, Flatten, TimeDistributed, Conv1D, MaxPooling1D

# data preparation


def download_usd():
    url = 'http://infra.datos.gob.ar/catalog/sspm/dataset/168/distribution/168.1/download/datos-tipo-cambio-usd-futuro-dolar-frecuencia-diaria.csv'
    r = requests.get(url)
    with open('dolar.csv', 'wb') as f:
        f.write(r.content)


def split_sequence(sequence, n_steps):
    """
    split a univariate sequence into samples
    """
    X, y = list(), list()
    for i in range(len(sequence)):
        # find the end of this pattern
        end_ix = i + n_steps
        # check if we are beyond the sequence
        if end_ix > len(sequence)-1:
            break
        # gather input and output parts of the pattern
        seq_x, seq_y = sequence[i:end_ix], sequence[end_ix]
        X.append(seq_x)
        y.append(seq_y)
    return numpy.array(X), numpy.array(y)

# download most recent data from datos.gob.ar
# download_usd()


data = pd.read_csv("dolar.csv")
historic_data = data.tipo_cambio_a3500.values[:-20]

# choose a number of time steps
n_steps = 4

# split into samples
X, y = split_sequence(historic_data, n_steps)

n_features = 1
n_seq = 2
n_steps = 2
n_epochs = 500
X = X.reshape((X.shape[0], n_seq, n_steps, n_features))

# model definition
model = Sequential()
model.add(TimeDistributed(Conv1D(filters=64,
                                 kernel_size=1,
                                 activation='relu'), input_shape=(None, n_steps, n_features)))
model.add(TimeDistributed(MaxPooling1D(pool_size=2)))
model.add(TimeDistributed(Flatten()))
model.add(LSTM(50, activation='relu'))  # we do no normalization, so use relu
model.add(Dense(1))
model.compile(optimizer='adam', loss='mse')

# model training
model.fit(X, y, epochs=n_epochs, verbose=0)

# predict dolar 🐬
predicted_values = []
last_predicted = data.tipo_cambio_a3500.values[-4:]
n_values_to_predict = 4

for x in range(n_values_to_predict):
    x_input = last_predicted.reshape((1, n_seq, n_steps, n_features))
    y = model.predict(x_input, verbose=0)
    y_predicted = y.flatten()[0]
    predicted_values.append(y_predicted)
    # calculate next batch of samples for prediction,
    # including the last prediction we just did.
    last_predicted = last_predicted[1:]
    last_predicted = numpy.append(last_predicted, [y])

print("Predicted values: " + str(predicted_values))


pyplot.title("Historico")
pyplot.plot(data.tipo_cambio_a3500)
pyplot.show()

pyplot.title("Prediccion " + str(n_values_to_predict) + " Días")
pyplot.plot(predicted_values)
pyplot.show()