In psHarmonize: Creates a Harmonized Dataset Based on a Set of Instructions

# RMarkdown child called by multi_source_compare.Rmd

# data_intermediate_mod is dataset already limited to visit of interest

# Use multi_source_vars_each character vector to determine source variables.

cont_source <- names(source_vars_type[source_vars_type == 'continuous'])
cont_source_index <- which(names(multi_source_vars_rename) %in% cont_source)
cont_source_rename <- multi_source_vars_rename[cont_source_index]

cat_source <- names(source_vars_type[source_vars_type == 'categorical'])
cat_source_index <- which(names(multi_source_vars_rename) %in% cat_source)
cat_source_rename <- multi_source_vars_rename[cat_source_index]

## Table ---------

for(i in seq_len(length(cont_source_rename)))
{

  cat('\n\n')
  cat(paste0('**Distribution of ', cont_source_rename[[i]],':**  \n\n'))
  cat('\n\n')

  table_summ <- data_intermediate_mod %>%
  group_by(cohort, visit) %>%
  summarise(n = n(),
            n_missing = sum(is.na(.data[[cont_source_rename[[i]]]])),
            min = min(.data[[cont_source_rename[[i]]]], na.rm = TRUE),
            Q1 = quantile(.data[[cont_source_rename[[i]]]], probs = c(0.25), na.rm = TRUE),
            median = median(.data[[cont_source_rename[[i]]]], na.rm = TRUE),
            Q3 = quantile(.data[[cont_source_rename[[i]]]], probs = c(0.75), na.rm = TRUE),
            max = max(.data[[cont_source_rename[[i]]]], na.rm = TRUE),
            mean = mean(.data[[cont_source_rename[[i]]]], na.rm = TRUE),
            std = sd(.data[[cont_source_rename[[i]]]], na.rm = TRUE)) %>%
  ungroup() %>%
  mutate(n = format(n, big.mark = ','),
         n_missing = format(n_missing, big.mark = ','),
         min = round(min, digits = 1),
         Q1 = round(Q1, digits = 1),
         median = round(median, digits = 1),
         Q3 = round(Q3, digits = 1),
         max = round(max, digits = 1),
         mean = round(mean, digits = 1),
         std = round(std, digits = 1))

  kable(table_summ) %>%
    kable_styling() %>%
  cat()


}

cat('\n\n')


# Print harmonized categorical variable

count_ns <- tabyl(data_intermediate_mod, value, show_missing_levels = FALSE) %>%
  untabyl() %>%
  adorn_totals(where = 'row') %>%
  mutate(n = format(n, big.mark = ','))

## Adding percentages and formatting
table_output <- tabyl(data_intermediate_mod, value, show_missing_levels = FALSE) %>%
  untabyl() %>%
  adorn_percentages(denominator = 'col') %>%
  adorn_totals(where = 'row') %>%
  adorn_pct_formatting(digits = 1) %>%
  adorn_ns(ns = count_ns, position = 'front')

## Print table
table_output %>%
  kable() %>%
  kable_styling() %>%
  row_spec(row = nrow(table_output), bold = TRUE) %>%
  cat()

 cat('\n')
 cat('\n')



# Plot ----------

to_plot <- ggplot(data = data_intermediate_mod, aes(x = .data[[cont_source_rename[[1]]]], y = .data[[cont_source_rename[[2]]]], color = factor(.data[['value']]), shape = factor(.data[['value']]))) +
  geom_point() +
  facet_wrap(~ factor(.data[[cat_source_rename]])) +
  theme_bw()


 print(to_plot)

cat('\n')
cat('\n')