Closed mjskay closed 3 years ago
Should be fixed now, thanks!
library(distributional)
test with a normal since hdrs should be \~= hilos
x = dist_normal(0,1)
correct at 50%
hdr(x, 50)
#> <list_of<hilo>[1]>
#> [[1]]
#> <hilo[1]>
#> [1] [-0.6744929, 0.6744929]50
hilo(x, 50)
#> <hilo[1]>
#> [1] [-0.6744898, 0.6744898]50
inverted?
hdr(x, 95)
#> <list_of<hilo>[1]>
#> [[1]]
#> <hilo[1]>
#> [1] [-1.95194, 1.95194]95
hilo(x, 95)
#> <hilo[1]>
#> [1] [-1.959964, 1.959964]95
hdrcde::hdr(generate(x, 1e6)[[1]], prob = 95)
#> $hdr
#> [,1] [,2]
#> 95% -1.955661 1.958104
#>
#> $mode
#> [1] -0.0006906591
#>
#> $falpha
#> 5%
#> 0.06643554
hdr(x, 5)
#> <list_of<hilo>[1]>
#> [[1]]
#> <hilo[1]>
#> [1] [-0.06393336, 0.06393336]5
hilo(x, 5)
#> <hilo[1]>
#> [1] [-0.06270678, 0.06270678]5
hdrcde::hdr(generate(x, 1e6)[[1]], prob = 5)
#> $hdr
#> [,1] [,2]
#> 5% -0.1045826 0.02071713
#>
#> $mode
#> [1] -0.05506228
#>
#> $falpha
#> 95%
#> 0.3982145
Created on 2021-07-06 by the reprex package (v2.0.0)
While looking at hdr, I just noticed that the intervals sizes seemed oddly small. I think the
size
parameter might be inverted...?Example:
Created on 2021-07-05 by the reprex package (v2.0.0)