SDPSYN2: General SDP function.
In DanielBonnery/BigSyn: X

Description Usage Arguments Details Examples

General SDP function.

SDPSYN2(
  TtableA,
  asis = NULL,
  notpredictor = asis,
  nrep = 1,
  synparameters = NULL,
  Sparameters = Sparameters.default.f(ref.table = TtableA, asis = asis, notpredictor =
    notpredictor, preferredmethod = "ctree", defaultsynparameters =
    c(as.list(synparameters),
    eval(formals(Sparameters.default.f)$defaultsynparameters)[setdiff(names(formals(Sparameters.default.f)$defaultsynparameters),
    c("", names(synparameters)))])),
  STtableA = if (is.null(asis)) {     data.frame(.n = rep(nrep, each = nrow(TtableA))) }
    else {     plyr::ddply(data.frame(.n = nrep), ~.n, function(d) {         TtableA[asis]
        }) },
  fitmodelsavepath = NULL,
  treeplotsavefolder = NULL,
  samplereportsavepath = NULL,
  stepbystepsavepath = NULL,
  doparallel = TRUE,
  recode = NULL,
  saveeach = 200,
  randomfitorder = TRUE,
  fitonly = FALSE
)

`TtableA`	a dataframe to synthesize
`asis`	list of variable names from TtableA to keep as is (e.g. not to synthesize)
`notpredictor`	list of variable names which should not be used as predictors
`nrep`	number of synthetic replicates wanted
`synparameters`	general synthetisation paramters
`Sparameters`	a list, Specific (variable by variable) synthetisation parameters, splits ...
`STtableA`	a dataframe
`fitmodelsavepath`	a path where to save the fitted models
`treeplotsavefolder`	a path where to save the tree plots
`samplereportsavepath`	a path where to save the sampling report
`stepbystepsavepath`	a path where to backup the synthetised in case of a crash
`doparallel`	a boolean indicating whether sampling should be done in parallel for each repliacte
`recode`	: a vector of character strings or NULL, list of variables to be recoded
`saveeach`	an integer, indicating every how many variables a backup is done
`randomfitorder`	a boolean : fitting for each variable can be done in the order of appearance of each variables or at random
`fitonly`	a boolean, if TRUE, no sampling is done.

This function is doing both the fitting and the sampling.

data(TtableA,package="BigSyn")
ATtableA=augmentT_f(TtableA,variablesmax="AA.present",
                   variablespct="AA.cont1")
asis=NULL;notpredictor=asis;nrep=1;synparameters=NULL;
Sparameters=
 Sparameters.default.f(ref.table=TtableA,
                       asis=asis,
                       notpredictor=notpredictor,
                       preferredmethod="ctree",
                       defaultsynparameters=
                          c(as.list(synparameters),
                            eval(formals(Sparameters.default.f)$defaultsynparameters)[
                              setdiff(names(formals(Sparameters.default.f)$defaultsynparameters),
                              c("",names(synparameters)))]));
SATtableA=plyr::rdply(nrep,ATtableA[asis]);
samplereportsavepath=NULL;
stepbystepsavepath=NULL;
doparallel=FALSE;
recode=NULL;
randomfitorder=TRUE;
fitonly=FALSE;
fitmodelsavepath=tempdir()
treeplotsavefolder=tempdir()
sapply(list.files(tempdir(),full.names = TRUE  ),file.remove)
SATtableA<-SDPSYN2(ATtableA,asis=NULL,
                  fitmodelsavepath = fitmodelsavepath,
                  treeplotsavefolder=treeplotsavefolder)
todisplay<-grep("La_La_Lrn1",names(STtableA[[1]]),value=T);
STtableA[[1]][1:3,todisplay];TtableA[1:3,todisplay]
##############################################################
# Controling that AA.present_La=0=>AA.present_La_Lb=0 in synthetic data
library(BigSyn)
library(reshape2)
library(data.table)
data(TtableA,package="BigSyn")
variablepct="AA.cont1"
variablespct=variablepct
variablemax="AA.present"
variablesmax=variablemax
set.seed(1)
asis=c("id1a", "id1b")
                  fitmodelsavepath=NULL
                  treeplotsavefolder=NULL
                  samplereportsavepath=NULL
                  stepbystepsavepath=NULL
                  doparallel=TRUE
                  recode=NULL
                  saveeach=200
                  randomfitorder=TRUE
                  fitonly=FALSE
                  
variablemax="AA.present"
variablesmax=variablemax
variablepct="AA.cont1"
variablespct=variablepct
ATtableA<-augmentT_f(TtableA,
                    variablesmax=variablesmax,variablespct=variablespct)
TtableA<-ATtableA
STtableA<-ATtableA[asis]
Sparameters=Sparameters.default.f(
   ref.table=ATtableA,asis=c("id1a", "id1b"),
   notpredictor=NULL,
   preferredmethod="ctree",
   defaultsynparameters=
     eval(formals(Sparameters.default.f)$defaultsynparameters))
SATtableA<-BigSyn::SDPSYN2(ATtableA,asis=c("id1a", "id1b"))[[1]]
problems<-SATtableA$AA.present_Lb_La==1&SATtableA$AA.present_Lb==0
mean(problems)
Sparameters[["AA.present_Lb_La"]]
library(dplyr)
library(ggplot2)
xx<-function(x){
xxx<-x[sort(grep("present",names(x),value=TRUE))]
xxx[xxx==0]<-NA
StudyDataTools::ggplot_missing(xxx)}
xx(ATtableA)
xx(SATtableA)