Heat plot is wrong with thinning

[osorensen]

Seems like we're dividing by the nmc parameter provided in compute_options and not by the actual number of stored iterations. Thanks to Marta for discovering.

library(BayesMallows)
set.seed(1)
mod <- compute_mallows(
  data = setup_rank_data(potato_visual),
  compute_options = set_compute_options(nmc = 10000, burnin = 400)
  )

plot(mod, parameter = "rho", items = 12)

heat_plot(mod)

mod <- compute_mallows(
  data = setup_rank_data(potato_visual),
  compute_options = set_compute_options(nmc = 10000, burnin = 400, rho_thinning = 10)
)

plot(mod, parameter = "rho", items = 12)

heat_plot(mod)

^{Created on 2024-02-16 with reprex v2.1.0}

[osorensen]

Good now

library(BayesMallows)
set.seed(1)
mod <- compute_mallows(
  data = setup_rank_data(potato_visual),
  compute_options = set_compute_options(nmc = 10000, burnin = 400)
  )

plot(mod, parameter = "rho", items = 12)

heat_plot(mod)

mod <- compute_mallows(
  data = setup_rank_data(potato_visual),
  compute_options = set_compute_options(nmc = 10000, burnin = 400, rho_thinning = 10)
)

plot(mod, parameter = "rho", items = 12)

heat_plot(mod)

^{Created on 2024-02-16 with reprex v2.1.0}

[wleoncio]

[osorensen]

Look at the probabilities legend at the lower plot

[osorensen]

Before

After

[wleoncio]

Oh good, thanks for pointing that out. I was obsessing with the plots and totally forgot to look at the legends.