nptvcm: NonParametric Time-Varying Coefficient Mmodel for Panel Data
In feiyoung/nptvcmPD: NonParametric Time-Varying Coefficient Mmodel for Panel Data

Description Usage Arguments Details Value Note Author(s) References See Also Examples

Nonparametric time-varying coefﬁcients model based penalty for longitudinal data with pre-speciﬁed ﬁnite time points, also known as panel data.

1	nptvcm (Xit, Yit, lambda=NULL, tuning.select='CV', nfolds=10, estimateSE=F, penalty='SCAD',nlambda=50, lambda.max=5)

`Xit`	a 3-dim n x T x p array, the observational value of covariates on panel data, the 1st index is sample index, the 2nd index is time index, the 3rd variable.
`Yit`	a n x T matrix, the observational value of response variable on panel data.
`lambda`	a positive value, user-specified penalty parameter, (optional) default as NULL.
`tuning.select`	a string, the method to select the penalty parameter, "CV" and "BIC" are supported in current version, (optional) default as "CV".
`nfolds`	a positive integer, specify the folds of cross validation when tuning.select='CV', (optional) default as 10.
`estimateSE`	a logical value, specify whether estimate the SE of estimated curve.
`penalty`	a string, specify the penalty function, supporting "SCAD", "MCP" and "laso", (optional) default as "SCAD".
`nlambda`	a positive integer, specify the grid of candidates on lambda, (optional) default as 50.
`lambda.max`	a positive value, specify the maximum value of lambda grids. (optional) default as 5.

For specific model and methodology, see reference.

return a correlation matrix with a type of specified structure.

nothing

Wei Liu

Huazhen Lin, Hyokyoung G. Hong, Baoying Yang, Wei Liu, Yong Zhang, Gang-Zhi Fan, Yi Li (2019). Nonparametric time-varying coeﬃcient models for panel data: Study of collection rate of public pension contributions. Statistics in Biosciences.

plot.nptvcm

  N=100;TT=5;
  datalist <- gendata(N, TT, 1)
  Yit <- datalist$Yit
  Xit <- datalist$Xit
  res <- nptvcm(Xit, Yit, estimateSE = T, tuning.select='CV')