lucashadfield / forecastNN

tools to easily build seq2seq and wavenet models on time series data

forecastNN

Tools to easily create and train a seq2seq model on a pandas DataFrame of data.

Initiate and Fit

from forecastNN.seq2seq import  Seq2SeqForecaster

model = Seq2SeqForecaster(pred_steps=10)
model.compile()
model.fit(train_df, epochs=20)

Predict

model.predict(test_df)

Plot Examples

model.plot_example_fit(test_df, loc=0) #Seq2SeqForecaster

model.plot_example_fit(test_df, loc=0, nstd=2) #Seq2SeqMonteCarloForecaster