select: select()

In jtkwar/GA: Genetic Algorithm for Variable Selection in Regression Problems

Description

This is main call function to run package GA. This package is comprised of a main execution file (select.R and a R file comtaining all functions that are necessary for execution (utils.R). The user can enter in a dataset and provide variables (listed below) to execute the genetic algorithm.

Usage

select(df, P, max_iter, method_text, fitness_function_text, mu,
crossover_points, mechanism, random = TRUE, Gap = 1/4, plot.return = FALSE)

Arguments

df	(data frame) Dataset to fit
P	(int) The number of individuals per generation.
max_iter	(int) The maximum number of iterations allowed when running GA
method_text	lm() or glm() methods for fitting the data
fitness_function_text	AIC, BIC, ow
random	(logical) Random replacement on or off for parent selection
Gap	Generation gap that determines how parents (generation t) are replaced by offspring of the (t+1) generation
plot.return	(logical) Boolean for returning plot at end of the algorithm
mu:	Mutation rate of the for each allele within the a given candidate chromosome.
crossover_points:	(int) The number of crossover points during breeding step
mechanism:	The mechanism to selection parents by. Selection mechanisms are "ranked" or "tournament."

Details

Contained in the list below are the invdividual functions that are called during the execution of the genetic algorithm.

• init(): Initializes the dataframe for the dataset

• training(): fits the method on candidates and return the fitness value of the candidate

• select_parents(): Chooses parents for breeding based off of ranked or tournament selection using the fitness values of each parent

• breed(): Breeds parents based off of pairing from select_parents()

• crossover(): Function within breed() to conduct crossover between parent pairs during breeding step

• mutation(): Determines if alleles of offspring (t+1 generation) mutates or not

• get_model(): Returns the best fit model of the dataset

jtkwar/GA documentation built on May 28, 2019, 7:36 a.m.