Generate features from a list of environmental variable vectors

Generates all configured feature types (linear, quadratic, product, threshold, hinge) from the supplied data vectors. Categorical variables are expanded into binary indicator features (one per distinct level) instead of the continuous feature types.

Usage

generate_features(
  data_list,
  feature_types = as.character(c("linear", "quadratic", "product", "threshold", "hinge")),
  n_thresholds = 10L,
  n_hinges = 10L,
  categorical_indices = as.integer(c())
)

Arguments

data_list

Named list of numeric vectors, one per environmental variable

feature_types

Character vector of feature types to generate. Valid values: "linear", "quadratic", "product", "threshold", "hinge". Defaults to all types.

n_thresholds

Number of threshold knots per variable (default: 10)

n_hinges

Number of hinge knots per variable (default: 10)

categorical_indices

Integer vector of 0-based indices identifying which variables in data_list are categorical (default: empty).

Value

List of external pointers to Feature objects