ragged_interval_predict: Get predictions plus uncertainty intervals on artificial...

In dhopp1/nowcastLSTM: LSTMs for Nowcasting

Get predictions plus uncertainty intervals on artificial vintages

Usage

ragged_interval_predict(
  model,
  pub_lags,
  lag,
  data,
  interval = 0.95,
  start_date = NULL,
  end_date = NULL,
  data_availability_weight_scheme = "fc"
)

Arguments

model	a trained LSTM model gotten from calling LSTM()
pub_lags	list of integers, list of periods back each input variable is set to missing. I.e. publication lag of the variable.
lag	integer, simulated periods back. E.g. -2 = simulating data as it would have been 2 months before target period, 1 = 1 month after, etc.
data	dataframe with the same columns the model was trained on
interval	number between 0 and 1, uncertainty interval. A closer number to one gives a higher uncertainty interval. E.g., 0.95 (95 \item start_datestring in "YYYY-MM-DD" format, start date of generating predictions. To save calculation time, i.e. just calculating after testing date instead of all dates \item end_datestring in "YYYY-MM-DD" format, end date of generating predictions \item data_availability_weight_schemestring, weighting scheme for data avilability. "fc" for weighting variables by feature contribution, "equal" for weighting each equally.

dataframe with periods, actuals if available, point predictions, lower and upper uncertainty intervals Get predictions plus uncertainty intervals on artificial vintages

dhopp1/nowcastLSTM documentation built on May 7, 2024, 9:30 a.m.