The MakeMatrix4TS object provides methods for making/building matrices from lagged endogenous and exogenous variables.
Attributes
- Companion
- trait
- Graph
-
- Supertypes
-
class Objecttrait Matchableclass Any
- Self type
-
MakeMatrix4TS.type
Members list
Value members
Concrete methods
Backfill the zero prefix of exogenous variable j (xej) by backcasting. The zero prefix will be at least of size 1 as 0.0 is initially prepended.
Backfill the zero prefix of exogenous variable j (xej) by backcasting. The zero prefix will be at least of size 1 as 0.0 is initially prepended.
Value parameters
- xej
-
the j-th exogenous variable vector
Attributes
Form an array of names for the features included in the model. Handles all *ARY* models. The *ARX* models require custom formNames methods.
Form an array of names for the features included in the model. Handles all *ARY* models. The *ARX* models require custom formNames methods.
Value parameters
- p
-
the number of lags for the endogenous variable (lags 1 to p)
- ps
-
the number of seasonal lags for the endogenous variable (lags 1 to ps)
- pwr
-
whether to raise the lagged endogenous values to a power (defaults to false)
- sp
-
the seasonal period (time units until repetitive behavior)
- spec
-
the number of trend terms
- start
-
the first seasonal lag to use (not subsumed by regular lags)
Attributes
Make/build a part of the input matrix consisting of the q * xe.dim2 columns for the exogenous variables.
Make/build a part of the input matrix consisting of the q * xe.dim2 columns for the exogenous variables.
Value parameters
- bakcast
-
whether a backcasted value is prepended to the time series
- q
-
the number of lags for each exogenous variable (lags 1 to q)
- qp
-
the power to raise the exogensous lags to
- xe
-
the matrix of exogenous variable values
Attributes
Given a response vector (time series) y, make/build and return an input/predictor MATRIX x for the linear LAG terms.
Given a response vector (time series) y, make/build and return an input/predictor MATRIX x for the linear LAG terms.
Value parameters
- bakcast
-
whether a backcasted value is prepended to the time series
- p
-
the maximum lag included (inclusive)
- y
-
the given output/response vector
Attributes
Given a response vector (time series) y, make/build and return an input/predictor MATRIX x for the linear LAG terms.
Given a response vector (time series) y, make/build and return an input/predictor MATRIX x for the linear LAG terms.
Value parameters
- bakcast
-
whether a backcasted value is prepended to the time series
- p
-
the maximum lag included (inclusive)
- y
-
the given output/response vector
Attributes
Given a response vector (time series) y, make/build and return an input/predictor MATRIX x for the SEASONALLY lagged terms.
Given a response vector (time series) y, make/build and return an input/predictor MATRIX x for the SEASONALLY lagged terms.
Value parameters
- bakcast
-
whether a backcasted value is prepended to the time series
- p
-
the maximum lag included (inclusive)
- ps
-
the number of seasonal lags
- sp
-
the seasonal period (time units until repetitive behavior)
- y
-
the given output/response vector
Attributes
Given a response vector (time series) y, make/build and return an input/predictor MATRIX x for the TREND terms.
Given a response vector (time series) y, make/build and return an input/predictor MATRIX x for the TREND terms.
Value parameters
- bakcast
-
whether a backcasted value is prepended to the time series
- lwave
-
the wavelength (distance between peaks)
- spec
-
the number of trend terms (added columns) 0 - none, 1 - constant 2 - linear, 3 - quadratic, 4 - sine, 5 - cosine
- y
-
the given output/response vector
Attributes
Given a response vector (time series) y, make/build and return an output/response MATRIX yy for each horizon to be forecasted (needed for DIRECT forecasting).
Given a response vector (time series) y, make/build and return an output/response MATRIX yy for each horizon to be forecasted (needed for DIRECT forecasting).
Value parameters
- bakcast
-
whether a backcasted value is prepended to the time series
- hh
-
the maximum forecasting horizon (h = 1 .. hh)
- y
-
the given output/response vector
Attributes
Concrete fields
Base hyper-parameter specification for regression based time series models.
Base hyper-parameter specification for regression based time series models.