lca-enumeration.Rmd

---
title: "Lab 1 - Latent Class Analysis Enumeration in MplusAutomation"
author: "Adam Garber"
date: "`r format(Sys.time(), '%B %d, %Y')`"
output:
  html_document: default
  pdf_document: default
subtitle: ""
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, eval = FALSE, message = FALSE, warning = FALSE, tidy = TRUE)

```

### **Note:** All models are estimated in Mplus via the interface R package {`MplusAutomation`}.

## ____________________________________

### Getting started:

1. Create an R-Project
2. Install packages
3. Load packages

### R-Project instructions:

1. click "NEW PROJECT" (upper right corner of window)
2. choose option "NEW DIRECTORY"
3. choose location of project (avoid long filepaths)

Within R-studio under the files pane (bottom right):

1. click "New Folder" and name folder "data"
2. click "New Folder" and name folder "enum_mplus"
3. click "New Folder" and name folder "figures"

### loading packages...
```{r}
library(MplusAutomation)
library(tidyverse)
library(relimp)
library(texreg)
library(rhdf5)
library(here)
library(glue)
library(kableExtra)
```

### filepaths are set using R-Projects & here::here()

### read in data 
```{r}

example_data <- read_csv(here("data","example_data.csv"),
                     na=c("",".", "999"))
```

## ____________________________________

## Enumeration (compare *k*-class models 1 - 6)

## ____________________________________

####Example has 6 indicator (x1 x2 x3 x4 x5)
```{r}

lca_k1_6  <- lapply(1:6, function(k) {
  lca_enum  <- mplusObject(
      
    TITLE = glue("C{k}_LCA_enumerate"), 
  
    VARIABLE = 
  glue(
    "categorical = x1 x2 x3 x4 x5; 
     usevar = x1 x2 x3 x4 x5;
    
     classes = c({k});"),
  
  ANALYSIS = 
   "estimator = mlr; 
    type = mixture; 
    starts = 500 100;",
  
  MODEL = "",
  OUTPUT = "",
  
  PLOT = 
    "type = plot3; 
    series = x1 x2 x3 x4 x5(*);",
  
  usevariables = colnames(ies_data),
  rdata = ies_data)

lca_enum_fit <- mplusModeler(lca_enum, 
                            dataout=glue(here("enum_mplus", "c_lca_enumerate.dat")),
                            modelout=glue(here("enum_mplus", "c{k}_lca_enumerate.inp")) ,
                            check=TRUE, run = TRUE, hashfilename = FALSE)
})

```

#### compare model fit for enumerated models
```{r}

output_enum <- readModels(here("enum_mplus"))


enum_summary <- LatexSummaryTable(output_enum, 
                keepCols=c("Title", 
                           "LL", 
                           "BIC",
                           "aBIC"), 
                                   sortBy = "Title")

enum_summary %>% 
  kable(booktabs = T, linesep = "") %>% 
  kable_styling(c("striped"), 
                full_width = F,
                position = "left")
```

## ____________________________________

### References: 

Hallquist, M. N., & Wiley, J. F. (2018). MplusAutomation: An R Package for Facilitating Large-Scale Latent Variable Analyses in Mplus. Structural equation modeling: a multidisciplinary journal, 25(4), 621-638.

Muthén, L.K. and Muthén, B.O. (1998-2017).  Mplus User’s Guide.  Eighth Edition. Los Angeles, CA: Muthén & Muthén

## ____________________________________