scripts_ejecucion/script_principal_ga.R

In papabloblo/sudoku: SUDOKU - Práctica 1 de Computación Evolutiva

# Datos de entrada --------------------------------------------------------
source("sudoku_prueba.R")

# Determinación de parámetros ---------------------------------------------

# Random Search -----------------------------------------------------------

set.seed(1234)
n <- 1

# tam_poblacion <- sample(10:200, size = n, replace = TRUE)
tam_poblacion <- rep(10, n)
prob_cruce <- runif(n = n, min = 0.5, max = 1)
tam_torneo <- sample(2:10, size = n, replace =TRUE)
# prob_mutacion <- runif(n = n, min = 0, max = 0.25)
prob_mutacion <- seq(from = 0, to = 5/81, length.out = n)

grid <- data.frame(tam_poblacion,
                   prob_cruce,
                   tam_torneo,
                   prob_mutacion
                   )

# Ejecución de pruebas ----------------------------------------------------

library(parallel)

system.time(resultados_facil <- mapply(pruebas_ga,
                                       num_pruebas              = 10,


                                       num_padres               = 2,

                                       tam_poblacion            = grid$tam_poblacion,
                                       prob_cruce               = grid$prob_cruce,
                                       tam_torneo               = grid$tam_torneo,
                                       prob_mutacion            = grid$prob_mutacion,

                                       max_iter                 = 5000,
                                       print_each               = 250,


                                       MoreArgs = list(generacion_poblacion_ini = generacion_poblacion,
                                                       genes_fijos              = sudoku_facil,
                                                       funcion_fitness          = fitness_sudoku,

                                                       valores_posibles         = 1:9,
                                                       valores_mutacion         = 1:9
                                                      )
                                       ,
                                       SIMPLIFY = FALSE
                                       ))

# save(resultados_facil, file = paste0("data/sudoku_facil_", Sys.Date(), ".RData"))

generar_data_frame <- function(resultado){
  resultado <- lapply(resultado, function(x){
    do.call(cbind, x)
  })


  resultado <- lapply(seq_along(resultado),
               function(x){
                 resultado[[x]]$id <- x
                 resultado[[x]]
               }
  )

  return(do.call(rbind, resultado))
}

resultados_dificil <- do.call(rbind,
                            lapply(resultados_dificil,
                                   generar_data_frame))

save(resultados_dificil, file = "data/sudoku_dificil.RData")
#
# library(tidyverse)
# library(ggthemes)
# resultados_facil <- tbl_df(resultados_facil)
#
# resultados_facil <- resultados_facil %>%
#   mutate(clave = paste0(parametros.tam_poblacion,
#                         parametros.prob_cruce,
#                         parametros.prob_mutacion,
#                         parametros.tam_torneo),
#          clave2 = paste0(clave, id))
#
# library(tidyverse)
# library(ggthemes)
#
# # Última iteración
# ultim_iter <- resultados_facil %>%
#   group_by(clave2) %>%
#   top_n(-1, wt = fitness) %>%
#   top_n(-1, wt = iteracion) %>%
#   ungroup()
#
# analisis_prob_mut <- ultim_iter %>%
#   group_by(clave,
#            parametros.prob_mutacion) %>%
#   summarise(fit_media  = mean(fitness),
#             fit_sd     = sd(fitness),
#             iter_media = mean(iteracion),
#             iter_sd    =  sd(iteracion)) %>%
#   arrange(fit_media)
#
#
# analisis_prob_mut %>%
#   ggplot(aes(x = parametros.prob_mutacion,
#              y = fit_media)) +
#     geom_point(aes(size = iter_media)) +
#     geom_smooth()
#
# analisis_prob_mut %>%
#   ggplot(aes(x = parametros.prob_mutacion,
#              y = fit_media)) +
#   geom_point(aes(size = fit_sd),
#              alpha = 0.7,
#              color = "steelblue") +
#   geom_smooth(colour = "firebrick",
#               method = "loess",
#               alpha = 0,
#               size = 2) +
#   theme_fivethirtyeight() +
#   ggtitle("Probabilidad de mutación vs. valor de fitness")
#
#
#
# analisis_prob_mut %>%
#   ggplot(aes(x = parametros.prob_mutacion,
#              y = iter_media)) +
#   geom_point(aes(size = iter_sd),
#              alpha = 0.7,
#              color = "steelblue") +
#   geom_smooth(colour = "firebrick",
#               method = "loess",
#               alpha = 0,
#               size = 2) +
#   theme_fivethirtyeight() +
#   ggtitle("Probabilidad de mutación vs. iteraciones")
#
#
#
#
# analisis_prob_mut %>%
#   ggplot(aes(x = iter_media)) +
#   geom_density(
#              alpha = 0.7,
#              color = "steelblue") +
#   theme_fivethirtyeight() +
#   ggtitle("Probabilidad de mutación vs. iteraciones")
#
#
# analisis_prob_mut %>%
#   ggplot(aes(x = iter_media,
#              group = cut_number(parametros.prob_mutacion, 2),
#              fill = cut_number(parametros.prob_mutacion, 2))) +
#   geom_density(
#     alpha = 0.7,
#     color = "steelblue") +
#   theme_fivethirtyeight() +
#   ggtitle("Probabilidad de mutación vs. iteraciones")
#
#
#
# analisis_prob_mut %>%
#   ggplot(aes(x = parametros.prob_mutacion,
#              y = fit_media,
#              group = cut_number(parametros.prob_mutacion, 30)))+
# geom_violin(scale = "width") +
#   geom_point()
#
# ultim_iter %>%
#   ggplot(aes(x = factor(parametros.prob_mutacion),
#              y = fitness)) +
#   geom_boxplot(alpha = 0) +
#   theme_fivethirtyeight()
#
#
#
# analisis_prob_mut %>%
#   ggplot() +
#   geom_point(aes(x = parametros.prob_mutacion,
#                  y = iter_media))
#
# resultados_facil2 <- resultados_facil[1:55000,]
# resultados_facil %>%
#   ggplot(aes(x = iteracion,
#              y = fitness)) +
#   geom_line(aes(group = clave2),
#             alpha = 0.5) +
#   geom_smooth(size = 2, color = "firebrick") +
#   theme_fivethirtyeight()
#
#
# resultados_facil %>%
#   ggplot(aes(x = iteracion,
#              y = fitness)) +
#   # geom_line(aes(group = clave2),
#   #           alpha = 0.5) +
#   geom_smooth(aes(group = clave),
#               alpha = 0.1,
#               color = "black",
#               size = 0.5) +
#   ylim(c(0, 75)) +
#   theme_fivethirtyeight()
#
#
#
#
# # Análisis de resultados --------------------------------------------------
#
# res <- resultados_facil %>%
#   filter(iteracion == 5000) %>%
#   group_by(parametros.tam_poblacion,
#            parametros.prob_cruce,
#            parametros.prob_mutacion,
#            parametros.tam_torneo) %>%
#   summarise(fitness_media = mean(fitness)) %>%
#   ungroup() %>%
#   arrange(fitness_media)
#
#
# res <- resultados_facil %>%
#   filter(iteracion == 5000) %>%
#   group_by(clave) %>%
#   summarise(fitness_media = mean(fitness)) %>%
#   ungroup() %>%
#   arrange(fitness_media) %>%
#   top_n(n = -20, wt = fitness_media)
#
# res %>%
#   left_join(resultados_facil) %>%
#   ggplot(aes(x = iteracion,
#              y = fitness)) +
#   # geom_line(aes(group = clave2),
#   #           alpha = 0.5) +
#   geom_smooth(aes(group = clave),
#               alpha = 0.1,
#               color = "black",
#               size = 0.5) +
#   ylim(c(0, 75)) +
#   theme_fivethirtyeight()
#
# mejores_20 <- res %>%
#   left_join(resultados_facil)
#
#
# mejores_20 %>%
#   ggplot(aes(x = iteracion,
#              y = fitness)) +
#   geom_line(aes(group = clave2),
#             alpha = 0.25)
#   ylim(c(0, 75)) +
#   theme_fivethirtyeight()
#
#
# mejores_20 %>%
#   ggplot(aes(x = parametros.prob_mutacion,
#              y = fitness)) +
#   geom_jitter(height = 0) +
#   geom_smooth()
#
#
#
# mejores_20 %>%
#   ggplot(aes(x = factor(parametros.prob_mutacion),
#              y = fitness)) +
#   geom_violin()