R/ceff_create_parnames.R

In chkiefer/countEffects: Estimation of Average and Conditional Treatment Effects on Count outcomes

ceff_create_parnames <- function(obj){

  inp <- obj@input
  ng <- inp@ng ## number of treatment groups
  nz <- inp@nz ## number of z
  nk <- inp@nk ## number of unfolded categories of K

  sep <- ""
  ## longer parameter names for many groups and/or covariates
  if(ng>9 | nk>9 | nz>9){sep <- "_"}

  # create list for alpha, beta and gamma coefficients
  tmp <- expand.grid(z=0:nz, k=0:(nk-1), x=0:(ng-1))
  alphas <- with(tmp, array(paste0("a",x,sep,k,sep,z), dim=c(nz+1,nk,ng)))

  label.g.function <- "(K,Z)"
  label.covs <- ",K,Z"
  if(nk==1 & nz==0){label.g.function <- "()"; label.covs <- ""}
  if(nk>1 & nz==0){label.g.function <- "(K)"; label.covs <- ",K"}
  if(nk==1 & nz>0){label.g.function <- "(Z)"; label.covs <- ",Z"}

  label.Egx <- paste0("E[g",1:(ng-1),label.g.function,"]")

  pk <- paste0("Pk",0:(nk-1))
  px <- paste0("Px",0:(ng-1))
  if(nz>0){
    tmp <- expand.grid(z=1:nz, k=0:(nk-1), x=0:(ng-1))
    cellmeanz <- with(tmp, paste0("mz",x,sep,k,sep,z))
    cellvarz <- with(tmp, paste0("vz",x,sep,k,sep,z))
    cellskewz <- with(tmp, paste0("sz",x,sep,k,sep,z))

    if (nz > 1){
      tmp0 <- t(combn(1:nz, 2))
      tmp0 <- apply(tmp0, 1, paste0, collapse = "")

      tmp <- expand.grid(k=0:(nk-1), x=0:(ng-1), zz = tmp0)
      cellcovz <- with(tmp, paste0("cvz",x,sep,k,sep,zz))
    } else {cellcovz <- character()}


    meanz <- paste0("Ez",1:nz)
    tmp <- expand.grid(k=0:(nk-1), z=1:nz)
    Ezk <- with(tmp, paste0("Ez",z,"k",k))
  }else{
    cellmeanz <- cellvarz <- meanz <- Ezk <- cellcovz <- cellskewz <- character()
  }

  tmp <- expand.grid(g=1:(ng-1), x=0:(ng-1), k=0:(nk-1), gx=0:(ng-1))
  cellexpc <- paste0("cellexp","gx",tmp$x,"k",tmp$k,"gx",tmp$gx,"k",tmp$k)

  ## E[E(Y|X=x,K,Z)]
  tmp <- expand.grid(x=0:(ng-1))
  Eygx <- paste0("Ey","gx",tmp$x,"kz")

  ## E[E(Y|X=x,K,Z)|X=x0]
  tmp <- expand.grid(x=0:(ng-1), x0=0:(ng-1))
  Eygxgx <- paste0("Ey","gx",tmp$x,"kzgx",tmp$x0)



  groupw <- paste0("gw",inp@vlevels$cell)
  relfreq <- paste0("relfreq",inp@vlevels$cell)
  Egx <- paste0("Eg",1:(ng-1))
  adjmeans <- paste0("adjmean",0:(ng-1))

  ## P(K=k|X=x)
  Pkgx=paste0("Pk",0:(nk-1),"gx",rep(0:(ng-1), each=nk))

  ## P(X=x|K=k)
  Pxgk=paste0("Px",0:(ng-1),"gk",rep(0:(nk-1), each=ng))

  ## E(Z|X=x)
  Ezgx <- character()
  if(nz>0){Ezgx=paste0("Ez",1:nz,"gx",rep(0:(ng-1), each=nz))}

  ## E(Z|K=k)
  Ezgk <- character()
  if(nz>0){Ezgk=paste0("Ez",1:nz,"gk",rep(0:(nk-1), each=nz))}

  ## E(Z*K|X=x)
  Ezkgx <- character()
  if(nz>0 & nk>1){
    tmp <- expand.grid(z=1:nz, k=0:(nk-1), x=0:(ng-1))
    Ezkgx=paste0("Ez",tmp$z,"k",tmp$k,"gx",tmp$x)
  }

  ## E(gx|X=x)
  tmp <- expand.grid(g=1:(ng-1), x=0:(ng-1))
  Egxgx <- paste0("Eg",tmp$g,"gx",tmp$x)

  ## E(gx|K=k)
  tmp <- expand.grid(g=1:(ng-1), k=0:(nk-1))
  Egxgk <- paste0("Eg",tmp$g,"gk",tmp$k)

  ## E(gx|X=x,K=k)
  tmp <- expand.grid(g=1:(ng-1), x=0:(ng-1), k=0:(nk-1))
  Egxgxk <- paste0("Eg",tmp$g,"gx",tmp$x,"k",tmp$k)

  ## average effect continuous covariate Z
  AveEffZ <- character()
  if(nz>0){AveEffZ <- paste0("AveEffZ",1:nz)}

  res <- new("parnames",
             alphas=alphas,
             label.g.function=label.g.function,
             label.covs=label.covs,
             label.Egx=label.Egx,
             cellmeanz=cellmeanz,
             cellvarz=cellvarz,
             cellcovz=cellcovz,
             cellskewz=cellskewz,
             cellexpc=cellexpc,
             Eygx=Eygx,
             Eygxgx=Eygxgx,
             meanz=meanz,
             pk=pk,
             px=px,
             Ezk=Ezk,
             Pkgx=Pkgx,
             Pxgk=Pxgk,
             Ezgx=Ezgx,
             Ezgk=Ezgk,
             Ezkgx=Ezkgx,
             groupw=groupw,
             relfreq=relfreq,
             Egx=Egx,
             Egxgx=Egxgx,
             Egxgk=Egxgk,
             Egxgxk=Egxgxk,
             adjmeans=adjmeans,
             AveEffZ=AveEffZ
  )

  return(res)
}