Missing.Rmd

---
title: "Missing"
author: "Pedro S.R. Martins"
date: "`r Sys.Date()`"
output:
  html_document:
    highlight: textmate
    code_folding: show
---

Code inspired by this blog post https://www.r-bloggers.com/2023/01/imputation-in-r-top-3-ways-for-imputing-missing-data/




# Packages

```{r}
pacman::p_load(tidyverse, #data manipulation
               janitor,
               easystats,
               purrr,
               knitr,
               mice) 
options(knitr.kable.NA = '')
base <-readRDS("basemiss.rds")
```


# Imputing data

```{r}
items <- base %>% select(starts_with("dass"), -ends_with("total"))

imputed_pmm = complete(mice(items, method = "pmm"))
imputed_midastouch = complete(mice(items, method = "midastouch"))
imputed_sample = complete(mice(items, method = "sample"))
imputed_rf = complete(mice(items, method = "rf"))
imputed_polr <- complete(mice(items %>% 
  mutate(across(starts_with("dass"), ~as.ordered(.x))), 
  method = "polr"))

results <- list(original = items,
                pmm=imputed_pmm,
                midastouch=imputed_midastouch,
                sample=imputed_sample,
                rf=imputed_rf,
                pol=imputed_polr)


```



# Visualizing frequencies for each imputation method {.tabset .tabset-fade}


## Using a loop   
  
  
  
Using a loop may not be as nice to make a communication plot.

   
```{r}
for(i in 1:length(results)){
  print(results[[i]] %>% 
    pivot_longer(everything()) %>% 
    ggplot(aes(x=value))+
    geom_histogram(stat = "count")+
    facet_wrap(name~.)+
    theme_bw()+
    labs(title = names(results)[[i]]))
}
```
  
  
  
## All in one plot 
  
  
```{r fig.width=17}
data.frame(results) %>% 
  mutate(across(pol.dass21_01:pol.dass21_21, ~ as.numeric(.x) - 1)) %>% 
  pivot_longer(everything()) %>% 
  separate(col = name,
           into = c("data","item"),
           sep = ".dass21_") %>% 
  group_by(data, item) %>% 
  count(value) %>% 
  mutate(data = factor(data, levels = c("original",
                                        "pmm",
                                        "midastouch",
                                        "sample",
                                        "rf",
                                        "pol"))) %>% 
  ggplot(aes(x = value, y = n, fill = data))+
  geom_bar(stat = "identity", 
           position = position_dodge(.9))+
  scale_fill_social()+
  facet_wrap(item~.)+
  theme_bw()
  
```


# Checking results using means and SD


```{r}
data.frame(results) %>% 
  mutate(across(pol.dass21_01:pol.dass21_21, ~ as.numeric(.x) - 1)) %>% 
  pivot_longer(everything()) %>% 
  separate(col = name,
           into = c("data","item"),
           sep = ".dass21_") %>% 
  mutate(data = factor(data, levels = c("original",
                                        "pmm",
                                        "midastouch",
                                        "sample",
                                        "rf",
                                        "pol"))) %>% 
  group_by(data, item) %>%
  summarise(mean = mean(value, na.rm = T),
            sd = sd(value, na.rm = T),
            ) %>% ungroup() %>% 

  pivot_wider(id_cols = item,
              values_from = c(mean, sd), 
              names_vary = "fastest", 
              names_from = data) %>% kable(., digits = 2)
```




# A plot for means and SD's
  
   
   
   

```{r}
data.frame(results) %>% 
  mutate(across(pol.dass21_01:pol.dass21_21, ~ as.numeric(.x) - 1)) %>% 
  pivot_longer(everything()) %>% 
  separate(col = name,
           into = c("data","item"),
           sep = ".dass21_") %>% 
  group_by(data, item) %>%
  summarise(mean = mean(value, na.rm = T),
            sd = sd(value, na.rm = T),
            ) %>% ungroup() %>%
  mutate(data = factor(data, levels = c("original",
                                        "pmm",
                                        "midastouch",
                                        "sample",
                                        "rf",
                                        "pol"))) %>% 
  ggplot(aes(x = item,
             y = mean,
             color = data,
             group = data))+
  geom_point(position = position_dodge(.5), size = 2)+
  scale_color_social()+
  geom_errorbar(aes(ymin = mean - sd,
                    ymax = mean + sd),
                # color = "grey60", 
                width = .2, size = .5,
                position = position_dodge(.5))+
  theme_classic()+
  theme(legend.position = "bottom")
```