Fix dbarts class prediction order

[qiushiyan]

Fix #780

[simonpcouch]

Looks great!

A quick check that this prediction is level-independent:

library(parsnip)

d <- data.frame(
  x1 = c(rnorm(100, mean = 0), rnorm(100, mean = 5)),
  x2 = c(rnorm(100, mean = 0), rnorm(100, mean = 5)),
  y = c(rep("negative", 100), rep("positive", 100))
)

d$y <- factor(d$y, levels = c("negative", "positive"))

model <- bart(mode = "classification", engine = "dbarts")

model_fit <- fit(model, y ~ ., data=d)

test <- data.frame(
  x1 = c(0, 5),
  x2 = c(0, 5)
)

predict(model_fit, new_data = test, type = "prob")
#> # A tibble: 2 × 2
#>   .pred_negative .pred_positive
#>            <dbl>          <dbl>
#> 1        0.992            0.008
#> 2        0.00600          0.994

predict(model_fit, new_data = test, type = "class")
#> # A tibble: 2 × 1
#>   .pred_class
#>   <fct>      
#> 1 negative   
#> 2 positive

# re-order levels to ensure independence
d$y <- factor(d$y, levels = c("positive", "negative"))

model <- bart(mode = "classification", engine = "dbarts")

model_fit <- fit(model, y ~ ., data = d)

predict(model_fit, new_data = test, type = "prob")
#> # A tibble: 2 × 2
#>   .pred_positive .pred_negative
#>            <dbl>          <dbl>
#> 1        0.00500          0.995
#> 2        0.995            0.005

predict(model_fit, new_data = test, type = "class")
#> # A tibble: 2 × 1
#>   .pred_class
#>   <fct>      
#> 1 negative   
#> 2 positive

^{Created on 2022-08-15 by the reprex package (v2.0.1)}

[github-actions]

This pull request has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.