Plotting with exuber • exuber

The plotting API has changed in exuber 0.4.0

The new design offers full flexibility and customization to produce publication-ready plots. exuber used to plot graph individually in a list, and then you could modify each plot and arrange them into a single grob with ggarrange()(which now is defunct). However, newer versions of exuber focus on providing a faceted plot as it easier to change the aesthetics and themes.

Let’s start by simulating some data.

set.seed(123)
sims <- tibble(
  sim_psy1 = sim_psy1(100),
  sim_psy2 = sim_psy2(100),
  sim_evans = sim_blan(100),
  sim_blan = sim_evans(100),
) 

# Esimation
estimation <- radf(sims, lag = 1)
  
# Critical Values
crit_values <- radf_mc_cv(nrow(sims))

Same Appearance with New Features

The visual output of autoplot in v0.4.0 is exactly the same as before.

autoplot(estimation, crit_values)
#> Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
#> ℹ Please use `linewidth` instead.
#> ℹ The deprecated feature was likely used in the exuber package.
#>   Please report the issue at <https://github.com/kvasilopoulos/exuber/issues>.
#> This warning is displayed once per session.
#> Call `lifecycle::last_lifecycle_warnings()` to see where this warning was
#> generated.

However, almost all aspects of the plot can be easily changed.

Change color and theme

The custom color for autoplot are “blue and”red”, however the user can easily override this option with ggplot2::scale_color_manual.

autoplot(estimation, crit_values) +
  scale_color_manual(values = c("grey","black")) + 
  theme_classic()

Changed the shaded region with shade_opt

shade_opt allows the user to manipulate the geom_rect() layer of the ggplot, using the shade function. Alternatively, it can be omitted if it set to NULL.

autoplot(estimation, crit_values, shade_opt = shade(fill = "pink", opacity = 0.3))

Custom plotting

Custom plotting is also very easy with the augment_join(), that merge the output of the estimation and and critical values in a ggplot2-friendly way.

joined <- augment_join(estimation, crit_values)
joined
#> # A tibble: 1,926 × 8
#>      key index id        data stat   tstat sig    crit
#>    <int> <dbl> <fct>    <dbl> <fct>  <dbl> <fct> <dbl>
#>  1    21    21 sim_psy1 119.  badf  -2.08  90    -0.44
#>  2    22    22 sim_psy1 112.  badf  -2.31  90    -0.44
#>  3    23    23 sim_psy1 111.  badf  -2.39  90    -0.44
#>  4    24    24 sim_psy1 104.  badf  -2.26  90    -0.44
#>  5    25    25 sim_psy1  98.6 badf  -2.08  90    -0.44
#>  6    26    26 sim_psy1  94.3 badf  -1.79  90    -0.44
#>  7    27    27 sim_psy1  82.9 badf  -1.00  90    -0.44
#>  8    28    28 sim_psy1  88.6 badf  -1.34  90    -0.44
#>  9    29    29 sim_psy1  89.6 badf  -1.28  90    -0.44
#> 10    30    30 sim_psy1  81.9 badf  -0.800 90    -0.44
#> # ℹ 1,916 more rows

The output of augment_join returns data in tidy format and offers full flexibility to the user. After this point plotting becomes extremely trivial.

joined %>% 
  ggplot(aes(x = index)) +
  geom_line(aes(y = tstat)) +
  geom_line(aes(y = crit)) +
  facet_grid(sig + stat ~  id  , scales = "free_y")
#> Warning: Removed 6 rows containing missing values or values outside the scale range
#> (`geom_line()`).
#> Removed 6 rows containing missing values or values outside the scale range
#> (`geom_line()`).

We also offer two functions scale_exuber_manual and theme_exuber that offer some extra functionality.

joined %>%
  pivot_longer(cols = c("tstat", "crit"), names_to = "nms") %>% 
  ggplot(aes(x = index, y = value, col = nms)) +
  geom_line() +
  facet_grid(sig + stat ~  id  , scales = "free_y") +
  scale_exuber_manual() +
  theme_exuber()

Distribution

In addition to critical values, we can also calculate the empirical distribution by utilizing the family of *_distr functions. For example if we can simulate the distribution of the supADF tests with Monte Carlo method.

distr <- radf_mc_distr(n = 300)
autoplot(distr)

Empirical distribution

This part is made just for fun.

library(tidyr)
distr %>%
  tidy() %>%
  rename_all(~ stringr::str_to_upper(.)) %>%
  gather(Statistic, value, factor_key = TRUE) %>%
  ggplot(aes(value, color = Statistic)) +
  stat_ecdf() +
  ggtitle("Empirical Cumulative Distribution") +
  geom_hline(yintercept = 0.95, linetype = "dashed") + theme_bw()

Old Functionality

To return to the old functionality there are several ways.

library(gridExtra)

# To choose only positive series (i.e. statistically significant for 5%)
positive_series <- diagnostics(estimation, crit_values)$positive 

# Through a loop on positive series 
plot_list1 <- list()
for (as in positive_series) {
  plot_list1[[as]] <- autoplot(estimation, crit_values, select_series = as)
}

# Alternatively  with lapply
plot_list2 <- lapply(positive_series, function(x) autoplot(estimation, crit_values, select_series = x))
names(plot_list2) <- positive_series

do.call(gridExtra::grid.arrange, plot_list1)

With the old functionality you had to make changes one at a time

plot_list1[[1]] <- plot_list1[[1]] + theme_classic()

and then reconstruct the plot with grid.arrange or some other function that arranges all plots into a single grob.

Enjoy Plotting with exuber !!!