Simple Weighted Averaging Ensemble

Alternative to stacking: weighted average of base model predictions. Simpler and more interpretable, often performs comparably.

Public fields

modalities

List of OmicModalitySpec objects

weights

Named numeric vector of modality weights

resampling

Resampling strategy for weight estimation

task_type

"classif" or "regr"

Methods

Public methods

• OmicWeightedEnsemble$new()

• OmicWeightedEnsemble$train()

• OmicWeightedEnsemble$predict()

• OmicWeightedEnsemble$print()

• OmicWeightedEnsemble$clone()

---

Method new()

Create a new weighted ensemble

Usage

OmicWeightedEnsemble$new(
  modalities,
  weights = NULL,
  resampling = NULL,
  task_type = "classif"
)

Arguments

modalities

List of OmicModalitySpec objects

weights

Optional named weights. If NULL, estimated from OOF performance.

resampling

Resampling strategy

task_type

"classif" or "regr"

Returns

An OmicWeightedEnsemble object

---

Method train()

Train the weighted ensemble

Usage

OmicWeightedEnsemble$train(
  y,
  target_name = "target",
  weight_method = c("inverse_logloss", "auc", "uniform")
)

Arguments

y

Target vector

target_name

Name for target column

weight_method

How to estimate weights: "inverse_logloss", "auc", "uniform"

Returns

Self (invisibly)

---

Method predict()

Predict using weighted averaging

Usage

OmicWeightedEnsemble$predict(newdata_list)

Arguments

newdata_list

Named list of data matrices

Returns

Data frame with weighted average predictions

---

Method print()

Print method

Usage

OmicWeightedEnsemble$print()

---

Method clone()

The objects of this class are cloneable with this method.

Usage

OmicWeightedEnsemble$clone(deep = FALSE)

Arguments

deep

Whether to make a deep clone.