LCT: Estimate a Latent Class Tree model
In MattisvdBergh/LCT: Estimate a Latent Class Tree model with Latent GOLD 5.1

Description Usage Arguments Details Value

Estimate a Latent Class Tree model with Latent GOLD 5.1

LCT(Dataset, LG, LGS = NULL, itemNames = NULL, mLevels = NULL,
  weight = "weight", resultsName = "", maxClassSplit1 = 2,
  maxClassSplit2 = 2, decreasing = TRUE, stopCriterium = "BIC",
  minSampleSize = 5, nKeepVariables = 0, namesKeepVariables = NULL,
  sets = 16, iterations = 50)

`Dataset`	A dataframe with the data or the filepath of the data.
`LG`	Filepath of the Latent GOLD executable
`LGS`	Filepath of Latent GOLD syntax for a model with 1- and 2-class splits
`itemNames`	The names of the indicators. If this is not given, all column names of the datafile will be used.
`mLevels`	A character vector being either "ordinal" or "continuous" to indicate the measurement level of each variable. It is required when LGS is specified.
`weight`	Name of the variable with the weights. When all records are unique observations, this should be one for every observation.
`resultsName`	Name of a folder which will be created in the working directory and contains all results by Latent GOLD.
`maxClassSplit1`	Maximum size of the first split of the tree. Will be assessed with the criterion given in stopCriterium. Defaults to two.
`maxClassSplit2`	Maximum size of each split after the first split of the tree. Defaults to two.
`decreasing`	Whether the ordering of classes should be decreasing or not. Defaults to TRUE.
`stopCriterium`	Criterium to decide on a split. Can be "LL" (logLikelihood), "AIC" or "BIC".
`minSampleSize`	Minimum sample size of a class. If this is below 1, a probability of the total sample size is used.
`nKeepVariables`	Number of variables to be kept if one wants to explore the results with external variables.
`namesKeepVariables`	Number of variables to be kept if one wants to explore the results with external variables.
`sets`	Name of the variable with the weights. When all records are unique observations, this should be one for every observation.
`iterations`	A character vector being either ordinal or continuous to indicate the measurement level of each variable. It is required when LGS is specified.

The LCT function constructs a LCT model by sequentially estimating 2-class models with Latent GOLD 5.1. This can be done automatically for standard models, but for more complex models a customized Latent GOLD syntax can be provided. The model size of the root can be increased with maxClassSplit1 and the remaining splits with maxClassSplit2.

Results of a Latent Class Tree analysis in an object of class 'LCT', which is a named list with two named lists. The first list contains information on the setup of the tree and the second list contains information on every split.

MattisvdBergh/LCT documentation built on May 8, 2019, 9:50 a.m.