Plotting interface

[nk027]

An interesting extension would be to facilitate using other packages for plotting etc -- primarily ggplot2, but maybe also ggdist, tidybayes or even broom.

This is part of the small project grant and an extension (possibly in the form of an additional package) should be implemented in the near future. For now I've played around with plotting IRF with ggplot2, which may be interesting:

gg_df <- function(x) {

  stopifnot(inherits(x, "bvar_irf"))

  # So we know what's what
  dimnames(x[["quants"]])[[2]] <- x[["variables"]]
  dimnames(x[["quants"]])[[4]] <- x[["variables"]]
  dimnames(x[["quants"]])[[3]] <- seq(dim(x[["quants"]])[3])

  out <- as.data.frame.table(x[["quants"]]) # Magic base R
  colnames(out) <- c("quant", "response", "time", "shock", "value")
  out[["time"]] <- as.integer(out[["time"]]) # Can't be a factor for the x axis

  return(out)
}

df <- gg_df(x)
df_ribbons <- tidyr::pivot_wider(df, names_from = quant)

ggplot(df, aes(x = time, y = value)) +
  facet_grid(response ~ shock) +
  geom_line(aes(color = quant))

ggplot(df_ribbons, aes(x = time)) + 
    facet_grid(response ~ shock) + 
    geom_line(aes(y = `50%`)) + 
    geom_line(aes(y = `16%`)) + 
    geom_line(aes(y = `84%`)) + 
    geom_ribbon(aes(ymin = `16%`, ymax = `84%`), alpha = 0.25)

[oDNAudio]

See BVARverse