Open wenjie2wang opened 3 months ago
Thanks for developing the package. I wonder how one should train a model with the pairwise loss. An example that reproduces the error is as follows:
## example of pairwise loss using {gbm3} library(gbm3) # gbm-developers/gbm3@a2b258c packageVersion("gbm3") #> [1] '3.0' train_x <- as.matrix(iris[, c("Sepal.Width", "Petal.Length", "Petal.Width")]) train_y <- iris$Sepal.Length ## train model <- gbm.fit( x = train_x, y = train_y, group = rep(1, length(train_y)), distribution = list(name = "pairwise", metric = "conc"), n.trees = 100, interaction.depth = 3, n.minobsinnode = 10, shrinkage = 0.1, bag.fraction = 0.5 ) #> Error in gbm_data_obj$x[train_params$id_order, , drop = FALSE]: incorrect number of dimensions
Created on 2024-03-28 with reprex v2.1.0
On another note related to the pairwise loss, I found unexpected prediction results from the pairwise loss by using the {gbm} package: https://github.com/gbm-developers/gbm/issues/79. It would be great if you could provide some insights. Thanks.
Thanks for developing the package. I wonder how one should train a model with the pairwise loss. An example that reproduces the error is as follows:
Created on 2024-03-28 with reprex v2.1.0
On another note related to the pairwise loss, I found unexpected prediction results from the pairwise loss by using the {gbm} package: https://github.com/gbm-developers/gbm/issues/79. It would be great if you could provide some insights. Thanks.