StockNN

Stock Prices Prediction Using Neural Network Models (Backpropagation, RNN LSTM, RBF) implemented in keras with Tensorflow backend to predict the daily closing price.

Class Version Usage

snn = stocknn().RNN()

snn = snn.preprocess('AAPL.csv', test_size=0.2)
snn = snn.train(batch_size=32, epochs=50)

or

snn = stocknn().BKP().preprocess('AAPL.csv', test_size=0.5).train(batch_size=16, epochs=25)

model = snn.save_model('AAPL')
mape = snn.test(model)[0]
pred = snn.predict(100)[1]

StockNN Subclasses

[Subclasses]:
    stocknn().RNN()            Recurrent Neural Networks.
    stocknn().RBF()            Radial Basis Function Networks.
    stocknn().BKP()            Back-propagation Networks.

Dataset

All datasets are obtained using pystocklib.

Requirement

• Keras
• Pandas
• numpy
• scikit-learn
• matplotlib

Credit

• PetraVidnerova