R/bind.R

In AnthonyTedde/vortex: The Machine Learning Vortex

bind <- function(data){
  list(
    crossbind = vortex::crossbind(data),
    rowbind = vortex::rowbind(data)
  )
}

crossbind <- function(data, ...){
  names(data) <- vortex::get_name(data)

  `%!in%` <- purrr::negate(`%in%`)
  if('by' %!in% names(list(...))){
    data <- data %>% purrr::map(tibble::rownames_to_column)
    return(vortex::crossbind(data, by = "rowname"))
  }

  #############################################################################
  ####
  ####  TODO: Fix Issue:
  ####   -> When multiple data.frame which cannot be conbined all together
  ####      The cross combination failed. A method to create cluster is needed.
  ####
  #############################################################################

  if(list(...)$by != "rowname"){
    fun <- get("map2", asNamespace("purrr"))
    data <- list(.x = data,
                 .y = names(data),
                 .f = ~tibble::rownames_to_column(
                   .x,
                   var = paste(.y, "rowname", sep = "."))) %>%
      do.call(what = fun, .)
  }

  bound_df <- Reduce(function(x, y) dplyr::inner_join(x, y, ...),
                     data)

  tryCatch(return(bound_df %>% tibble::column_to_rownames()),
    error = function(cond){
      return(bound_df)}
  )
}

rowbind <- function(data, ...){
  names(data) <- vortex::get_name(data)
  columns <- vortex::get_column_selection(data)

  newdataset_name <- purrr::map(columns, ~paste(.x$data.frame, collapse = "_"))

  purrr::map(columns, function(x){
    data[x$data.frame] %>% purrr::map(~.x[, x$columns]) %>%
      purrr::reduce(rbind)
  })%>%
      structure(names = newdataset_name)
}