convertSignatureToValueOutOfSample {patterncausality} | R Documentation |
This function predicts future values for a time series starting from the last known predicted value. It uses a series of predicted signature changes to extrapolate future values, assuming that changes continue in a similar manner to the signatures provided. This is particularly useful in scenarios where short-term predictions are made in complex systems based on embedded dynamic structures as described in the associated literature.
convertSignatureToValueOutOfSample(
E,
tau,
Y_pred_last,
i,
h,
predictedSignatureY
)
E |
Integer, the number of future values to predict. |
tau |
Integer, the time delay used in the system dynamics; while not used directly in this function, it is relevant in the broader context of the methodology for embedding time series data. |
Y_pred_last |
Numeric, the last predicted value from which future values will be extrapolated. |
i |
Integer, the starting index for the prediction; not used in this function but generally important in the broader algorithm. |
h |
Integer, the horizon step from the last known actual value to the first prediction point; not used directly in this function but part of the overall predictive framework. |
predictedSignatureY |
Numeric vector, the predicted incremental changes (signature) at each step used for extrapolation of the series. |
Numeric vector containing the extrapolated future values of the series, starting from Y_pred_last and extending for E steps, adjusted by the predicted signature changes.
# Suppose Y_pred_last is the last known predicted value of a financial time series
Y_pred_last <- 120
# Assume predicted signature changes based on a model's output
predictedSignatureY <- c(2, 3, 4, 5, 6)
# Number of future points to predict
E <- 5
# Example values for tau and i, h not used in this function directly
tau <- 1
i <- 1
h <- 1
# Generate future predictions
futureValues <- convertSignatureToValueOutOfSample(E, tau, Y_pred_last, i, h, predictedSignatureY)
print(futureValues)