1
$\begingroup$

TensorFlow 2 "Time series forecasting tutorial" (https://www.tensorflow.org/tutorials/structured_data/time_series#recurrent_neural_network) gives an example of a LSTM multi-step prediction model that given a past history predicts a range of future values:

# Prepare data
train_data_multi = tf.data.Dataset.from_tensor_slices((x_train_multi, y_train_multi))
train_data_multi = train_data_multi.cache().shuffle(BUFFER_SIZE).batch(BATCH_SIZE).repeat()

val_data_multi = tf.data.Dataset.from_tensor_slices((x_val_multi, y_val_multi))
val_data_multi = val_data_multi.batch(BATCH_SIZE).repeat()

# Model with two LSTM layers
multi_step_model = tf.keras.models.Sequential()
multi_step_model.add(tf.keras.layers.LSTM(32,
                                      return_sequences=True,
                                      input_shape=x_train_multi.shape[-2:]))
multi_step_model.add(tf.keras.layers.LSTM(16, activation='relu'))
multi_step_model.add(tf.keras.layers.Dense(72))
multi_step_model.compile(optimizer=tf.keras.optimizers.RMSprop(clipvalue=1.0), loss='mae')

# Train model
multi_step_history = multi_step_model.fit(train_data_multi, epochs=EPOCHS,
                                      steps_per_epoch=EVALUATION_INTERVAL,
                                      validation_data=val_data_multi,
                                      validation_steps=50)

Tutorial also shows how to plot some values predicted by LSTM:

def multi_step_plot(history, true_future, prediction):
     ...
     ...

for x, y in val_data_multi.take(3):
     multi_step_plot(x[0], y[0], multi_step_model.predict(x)[0])

Question: How to iterate all x-s and all y-s (true values) in a validation dataset (val_data_multi)? I can not figure out how to access these values in tf.data.Dataset.from_tensor_slices. Also, how to get predictions from the model for all entries in validation dataset to be next used in calculation of mean MAE and RMSE of the model?

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.