grf package options

Source: R/grf_options.R

grf package options can be set using R's options command. The current available options are:

  • `grf.verbose`: controls whether information is printed during training. The default value is `FALSE`.

  • `grf.legacy.seed`: controls whether grf's random seed behavior depends on the number of CPU threads used to train the forest. The default value is `FALSE`. Set to `TRUE` to recover results produced with grf versions prior to 2.4.0.

grf_options()

Value

Prints the current grf package options.

Examples

# \donttest{
# Enable progress output during training.
options(grf.verbose = TRUE)
n <- 1500
p <- 10
X <- matrix(rnorm(n * p), n, p)
Y <- X[, 1] * rnorm(n)
r.forest <- regression_forest(X, Y)
#> 
training [regression]:   0%|          | 0/2000 [00:00:00<?, ?it/s]
training [regression]:   0%|          | 0/2000 [00:00:00<?, ?it/s]
training [regression]:   0%|          | 0/2000 [00:00:00<?, ?it/s]
training [regression]:   0%|          | 0/2000 [00:00:00<?, ?it/s]
training [regression]:   7%|█         | 146/2000 [00:00:00<00:00:01, 1392.45it/s]
training [regression]:  15%|█         | 292/2000 [00:00:00<00:00:01, 1420.39it/s]
training [regression]:  22%|██        | 438/2000 [00:00:00<00:00:01, 1429.96it/s]
training [regression]:  29%|███       | 586/2000 [00:00:00<00:00:00, 1440.31it/s]
training [regression]:  37%|████      | 732/2000 [00:00:00<00:00:00, 1442.30it/s]
training [regression]:  44%|████      | 876/2000 [00:00:00<00:00:00, 1440.07it/s]
training [regression]:  51%|█████     | 1020/2000 [00:00:00<00:00:00, 1438.66it/s]
training [regression]:  58%|██████    | 1166/2000 [00:00:00<00:00:00, 1439.90it/s]
training [regression]:  66%|███████   | 1314/2000 [00:00:00<00:00:00, 1443.80it/s]
training [regression]:  73%|███████   | 1458/2000 [00:00:01<00:00:00, 1442.45it/s]
training [regression]:  80%|████████  | 1602/2000 [00:00:01<00:00:00, 1441.45it/s]
training [regression]:  87%|█████████ | 1748/2000 [00:00:01<00:00:00, 1442.05it/s]
training [regression]:  95%|█████████ | 1892/2000 [00:00:01<00:00:00, 1441.14it/s]
training [regression]: 100%|██████████| 2000/2000 [00:00:01<00:00:00, 1442.30it/s]
options(grf.verbose = FALSE)

# Print current package options.
grf_options()
#>     grf.verbose grf.legacy.seed 
#>           FALSE           FALSE 

# Use random seed behavior prior to version 2.4.0.
options(grf.legacy.seed = TRUE)

# Use random seed independent of num.threads (default as of version 2.4.0 and higher).
options(grf.legacy.seed = FALSE)
# }