Outcome / partial residual used to sample an additive model.

The outcome class is wrapper around a vector of (mutable) outcomes for ML tasks (supervised learning, causal inference). When an additive tree ensemble is sampled, the outcome used to sample a specific model term is the "partial residual" consisting of the outcome minus the predictions of every other model term (trees, group random effects, etc...).

Public fields

data_ptr

External pointer to a C++ Outcome class

Methods

Public methods

• Outcome$new()

• Outcome$get_data()

---

Method new()

Create a new Outcome object.

Usage

Outcome$new(outcome)

Arguments

outcome

Vector of outcome values

Returns

A new Outcome object.

---

Method get_data()

Extract raw data in R from the underlying C++ object

Usage

Outcome$get_data()

Returns

R vector containing (copy of) the values in Outcome object