badmf.fit: Bayesian Decision-Making Forest (BaD-MF)
In ebridge2/badmf: Bayesian Decision-Making Forests (BaD-MFs)

Description Usage Arguments Value Author(s)

Fit a Bayesian Decision-Making Forest with a stats-like formula frontend interface.

badmf.fit(formuler, data = NULL, d = NULL, alpha = NULL,
  ntrees = 10L, bagg = 0.632, method = "classification",
  depth.max = 1L, size = 1L, debug = FALSE, mc.cores = 1L,
  train.params = NULL, ...)

`formuler`	ravioli ravioli give me the formuoli.
`data`	the data associated with the formuler. Note: if you want an intercept, you must add it ahead of time.
`d`	the number of features to subsample at a split node. Defaults to `sqrt(nsamples)`.
`alpha`	the feature sampling prior. Should be a `[p]` vector, where `p` is the number of predictors. Corresponds to alpha for a Dirichlet distribution. If `NULL`, samples uniformly for the initial training iteration.
`ntrees`	the number of trees to construct. Defaults to `10L`.
`bagg`	the relative size of the subsamples for the training set. A numeric s.t. `0 < bagg <= 1`. Each subsample will be `bagg*n` elements. Defaults to `0.632`.
`depth.max`	the maximum allowed tree depth. Defaults to `5L`.
`size`	the minimum allowed number of samples for an individual node. Defaults to `1L`.
`debug`	whether to save the predictors and responses that are categorized. Defaults to `FALSE`.
`mc.cores`	the number of cores to use. Should be `0 < mc.cores <= parallel::detectCores()`. Defaults to `1L`.
`train.params`	if you wish to provide specialized parameters for training, a named list containing the following named elements: `d` the number of features to subsample at a split node. `ntrees` the number of trees to construct. `bagg` the relative size of the subsamples from the training set. `depth.max` the maximum allowed tree depth. `size` the minimum allowed number of samples for an individual node. Any unset parameters will default to the values provided above (or the corresponding defaults if unprovided).
`...`	trailing arguments.