Creates a population of index models.

This function is used in terga1 and generates a list of index vectors in the variable space. These vectors can be unique or not. NB that if clf$params$unique_vars is set to TRUE it can take a long time to come out of the while loop which ensures the uniqueness of the individuals.

Usage

population(
  clf,
  size_ind,
  size_world,
  best_ancestor = NULL,
  size_pop = NULL,
  seed = NULL
)

Arguments

clf:

the classifier parameter object

size_ind:

The sparsity of the models. All the models of this population will have the same number of features.

size_world:

The number of features from which we can choose the indices. This is needed to compute the combinatory space search.

best_ancestor:

We can supply to the popolution an individual (vector with indeces) of a lower sparsity. This will ensure to seed part of the population with at least those genes. We added this feature after an observations that a local optimum of lower sparsity was lost in higher sparsities.

size_pop:

the number of models to produce (default=NULL). This information is stored here clf$params$size_pop, but this parameter allows to override it.

Value

a population of index models