R/utils_tsl_pairs.R

In distantia: Advanced Toolset for Efficient Time Series Dissimilarity Analysis

#' Data Frame with Pairs of Time Series in Time Series Lists
#'
#' @description
#' Internal function used in [distantia()] and [momentum()] to generate a data frame with combinations of time series and function arguments.
#'
#'
#' @param tsl (required, list) Time series list. Default: NULL
#' @param args_list (required, list) arguments to combine with the pairs of time series. Default: NULL
#'
#' @return data frame
#' @export
#' @autoglobal
#' @family internal
utils_tsl_pairs <- function(
    tsl = NULL,
    args_list = NULL
){

  df_tsl <- data.frame(
    t(
      utils::combn(
        names(tsl),
        m = 2
      )
    )
  )

  names(df_tsl) <- c(
    "x",
    "y"
  )

  if(is.null(args_list)){
    return(df_tsl)
  }

  df_args <- expand.grid(
    args_list,
    stringsAsFactors = FALSE
  )

  df <- merge(
    x = df_tsl,
    y = df_args
  )

  #remove invalid combinations

  #remove invalid commbinations of permutation x block_size
  free <- c("free", "free_by_row")

  if(
    all(c("permutation", "block_size") %in% colnames(df)) &&
    any(free %in% df$permutation)
  ){

    df[df$permutation %in% free, "block_size"] <- NA

  }

  #remove invalid combinations of lock_step and bandwidth
  if(
    all(c("lock_step", "bandwidth") %in% colnames(df)) &&
    any(df$lock_step == TRUE)
  ){

    df[df$lock_step ==TRUE, "bandwidth"] <- NA

  }

  #remove invalid combinations of lock_step and diagonal
  if(
    all(c("lock_step", "diagonal") %in% colnames(df)) &&
    any(df$lock_step == TRUE)
  ){

    df[df$lock_step == TRUE, "diagonal"] <- NA

  }

  df <- unique(df)

  df


}