ARIMA_diff
The ARIMA_diff object provides methods for taking first and second order differences, as well as transforming back to the original scale.
diff: position y --> velocity v --> acceleration a (actual) | | | backform, undiff: position yp <-- velocity vp <-- acceleration ap (predicted)
Attributes
- See also
-
stats.stackexchange.com/questions/32634/difference-time-series-before-arima-or-within-arima
- Graph
-
- Supertypes
-
class Objecttrait Matchableclass Any
- Self type
-
ARIMA_diff.type
Members list
Value members
Concrete methods
Backform by transforming back the predicted values of a differenced time series to the original scale using actual values y. E.g., for d = 1: Velocity vp -> Position yp.
Backform by transforming back the predicted values of a differenced time series to the original scale using actual values y. E.g., for d = 1: Velocity vp -> Position yp.
Value parameters
- d
-
the order of simple differencing (defaults to 1)
- vp
-
the predicted differenced (velocity/acceleration) time series
- y
-
the actual position time series vector (first d values needed)
Attributes
Take the d'th difference of the position time series for d in {0, 1, 2}. A new vector (of length y.dim-d) is returned even when there is no difference taken (d = 0), to ensure the original is preserved. E.g., for d = 1: Position y --> Velocity v.
Take the d'th difference of the position time series for d in {0, 1, 2}. A new vector (of length y.dim-d) is returned even when there is no difference taken (d = 0), to ensure the original is preserved. E.g., for d = 1: Position y --> Velocity v.
Value parameters
- d
-
the order of simple differencing (defaults to 1)
- y
-
the actual position time series to be differenced
Attributes
Transform the forecast values in the FORECAST MATRIX of a differenced time series back to the original scale for all horizons (1 to h).
Transform the forecast values in the FORECAST MATRIX of a differenced time series back to the original scale for all horizons (1 to h).
Value parameters
- d
-
the order of simple differencing
- vf
-
the matrix of all multi-horizon forecasted values (differenced)
- y
-
the original actual time series vector (undifferenced)
Attributes
Transform the forecast values from time point t of a differenced time series back to the original scale for all horizons (1 to h).
Transform the forecast values from time point t of a differenced time series back to the original scale for all horizons (1 to h).
Value parameters
- d
-
the order of simple differencing
- t
-
the time point being forecasted (@see the
forecastmethod) - vh
-
the vector of forecasted differenced values for times t+1, ... t+h
- y
-
the original actual time series vector (undifference)
Attributes
Undifference the velocity time series by adding the difference to the previous value. Velocity v -> Position y.
Undifference the velocity time series by adding the difference to the previous value. Velocity v -> Position y.
Value parameters
- v
-
the differenced time series (velocity)
- y0
-
the first value in the original time series