todo-files/svetlana.R

In mlr-org/mlrMBO: Bayesian Optimization and Model-Based Optimization of Expensive Black-Box Functions

library(methods)
library(testthat)
library(devtools)
library(BBmisc)
library(mlr)
library(soobench)
library(mco)
library(checkmate)
library(ggplot2)
library(gridExtra)

load_all(".", reset=TRUE)

source("todo-files/plotOptPath.R")
source("todo-files/renderPCPPlot.R")

set.seed(6)

z =load2("~/Desktop/saveopt.RData")
print(plotOptPath(z$opt.path, z$control))


# ps = z$par.set
# opdf = as.data.frame(z$opt.path)
# xy = as.data.frame(z$opt.path, include.rest = FALSE)
# task = makeRegrTask(data = xy, target = "y")

# z$learner$fix.factors = FALSE
# mod = train(z$learner, task)
# des = generateRandomDesign(1000L, z$opt.path$par.set)
# pred = predict(mod, newdata = des)$data$response
# idx = pred < 0.43
# good = des[idx, ]
# r= sapply(good, range)
# print(r)