Create an object of the S3 class PSTR for later usage. This function should be run prior to the other functions in the package. It will return an object which you will use as an input for the other functions. It builds up the basic settings for the Panel Smooth Transition Regression (PSTR) Modelling.
a tibble of data. The number of rows of data must be the sample size iT times individuals number N.
dep
column number or name of the dependent variable.
indep
a vector of column numbers of names of the independent variables.
indep_k
a vector of column numbers of names of the independent variables in the nonlinear part. If indep_k is not given (= NULL), the nonlinear part will be the same as the linear part.
tvars
a vector of column numbers or names of the potential transition variables to be tested.
im
maximal number of switches in the transition function used in the linearity evaluation tests, by default im=1.
iT
sample size.
Value
An object of the class PSTR for later usage.
The object is a list containing the following components:
iT
the time length of the panel
iN
the number of individuals
vY
the vector of the dependent variable
mX
the matrix of the explanatory variables in the linear part
mK
the matrix of the explanatory variables in the nonlinear part
mQ
the matrix of the potential transition variables
im
the maximal number of switches used in the linearity test
Details
Potential transition variables in tvars will be tested one by one in, for example, LinTest function.
There is no need to specify the number of individuals, as it will be obtained automatically inside the function given the number of rows and the sample size iT.
NAs in data are removed automatically inside the function.