gafs.control: Genetic Algorithm Feature Selection Parameters

In dsnavega/cbfsr: cbfsr: Tools for correlation-based feature selection in R

Description

Genetic Algorithm Feature Selection Parameters

Usage

gafs.control(
  npopulation = 100,
  ngeneration = 200,
  pressure = 0.8,
  crossover = 0.8,
  mutation = 0.05,
  elitism = 0.05,
  warmup = 0.5,
  early = 0.05,
  adaptive = T,
  rate = 0.01,
  alpha = 0.01,
  digits = 3
)

Arguments

npopulation	number of individuals in the initial population.
ngeneration	number of generation
pressure	selective pressure probability. Probability of selecting the best individual in a binary tournament selection
crossover	crossover probability.
mutation	mutation probability.
elitism	fraction of best individuals that are transferred directly to the next generation.
warmup	fraction of generations at which early stop and adaptive behavior will trigger.
early	fraction of consecutive generations at which the GA will be terminated if no improvement is detected to the fitness function.
adaptive	a logical stating if probabilities defined by pressure, crossover and mutation should be adapted according to an exponential function after the warm-up phase. Default is TRUE.
rate	the rate of the exponential function to adapt the parameters of the GA. Default is 0.01 (i.e. probability will increase or decrease by 1%)
alpha	level of significance of binomial test performed to assess if a feature should be accepted or rejected based the the best individual chromosomes of all generations. Default is 0.01
digits	an integer defining the precision of the round function. Default is 3.

Author(s)

David Senhora Navega

dsnavega/cbfsr documentation built on Jan. 1, 2021, 12:07 a.m.