Time Series prediction using RNNs (Keras) in R

Question

I was following the Chollet's Deep learning with R approach ( fitting RNNs to time series data ) for fitting RNNs for time series prediction.

model <- keras_model_sequential() %>% 
  layer_gru(units = 32, 
            dropout = 0.1, 
            recurrent_dropout = 0.5,
            return_sequences = TRUE,
            input_shape = list(NULL, dim(data)[[-1]])) %>% 
  layer_gru(units = 64, activation = "relu",
            dropout = 0.1,
            recurrent_dropout = 0.5) %>% 
  layer_dense(units = 1)

model %>% compile(
  optimizer = optimizer_rmsprop(),
  loss = "mae"
)

history <- model %>% fit_generator(
  train_gen,
  steps_per_epoch = 500,
  epochs = 40,
  validation_data = val_gen,
  validation_steps = val_steps
)

Here, train, validation and tests data are generated using:

lookback <- 1440
step <- 6
delay <- 144
batch_size <- 128

train_gen <- generator(
  data,
  lookback = lookback,
  delay = delay,
  min_index = 1,
  max_index = 200000,
  shuffle = TRUE,
  step = step, 
  batch_size = batch_size
)

val_gen = generator(
  data,
  lookback = lookback,
  delay = delay,
  min_index = 200001,
  max_index = 300000,
  step = step,
  batch_size = batch_size
)

test_gen <- generator(
  data,
  lookback = lookback,
  delay = delay,
  min_index = 300001,
  max_index = NULL,
  step = step,
  batch_size = batch_size
)

# How many steps to draw from val_gen in order to see the entire validation set
val_steps <- (300000 - 200001 - lookback) / batch_size

# How many steps to draw from test_gen in order to see the entire test set
test_steps <- (nrow(data) - 300001 - lookback) / batch_size

After this, I have read the Keras documentation and found the prediction function. To find the prediction on test data:

m <- model %>% evaluate_generator(test_gen, steps = test_steps)
m

However, It giving only loss value for test data.

My question is, how to get the predictions for each point in test dataset, like we can get in other time series method? How to plot these predicted and actual values?

Answer 1

It seems to me that you need to redefine the generator , you need to get only the samples as output. Following your example:

# generator function
generator <- function(data, lookback, delay, min_index, max_index,
                      shuffle = FALSE, batch_size = 128, step = 6) {
  if (is.null(max_index))
    max_index <- nrow(data) - delay - 1
  i <- min_index + lookback
  function() {
    if (shuffle) {
      rows <- sample(c((min_index+lookback):max_index), size = batch_size)
    } else {
      if (i + batch_size >= max_index)
        i <<- min_index + lookback
      rows <- c(i:min(i+batch_size-1, max_index))
      i <<- i + length(rows)
    }

    samples <- array(0, dim = c(length(rows), 
                                lookback / step,
                                dim(data)[[-1]]))
    targets <- array(0, dim = c(length(rows)))

    for (j in 1:length(rows)) {
      indices <- seq(rows[[j]] - lookback, rows[[j]]-1, 
                     length.out = dim(samples)[[2]])
      samples[j,,] <- data[indices,]
      targets[[j]] <- data[rows[[j]] + delay,2]
    }            

    list(samples) # just the samples, (quick and dirty solution, I just removed targets)
  }
}

# test_gen is the same
test_gen <- generator(
  data,
  lookback = lookback,
  delay = delay,
  min_index = 300001,
  max_index = NULL,
  step = step,
  batch_size = batch_size
)

Now you can call predict_generator :

preds <- model %>% predict_generator(test_gen, steps = test_steps)

But now you need to de-normalize those, because you scaled each variable before the fit.

denorm_pred = preds * std + mean

Be careful that std and mean should be calculated on T (degC) only on the train data, otherwise you're overfitting.