tests/testthat/test-regressor.R

In thllwg/tpotr: An R-Wrapper for the Python Automated Machine Learning tool that optimizes machine learning pipelines using genetic programming (TPOT)

context("Regression with tpotr")

test_that("regression works in tpotr", {
  library("caTools")
  data_wine <- read.csv(url("https://archive.ics.uci.edu/ml/machine-learning-databases/wine-quality/winequality-red.csv"), header = TRUE, sep=";")
  data_wine$"quality" <- as.numeric(as.character(data_wine$"quality"))

  #Training-Test Split
  set.seed(101)
  sample_wine = sample.split(data_wine$`fixed.acidity`, SplitRatio = 2/3)
  train_wine = subset(data_wine, sample_wine == TRUE)
  train_wine.features = train_wine[,1:11]
  train_wine.classes = train_wine[,12]
  test_wine  = subset(data_wine, sample_wine == FALSE)
  test_wine.features = test_wine[,1:11]
  test_wine.classes = test_wine[,12]

  #Pipeline
  tpot <- TPOTRRegressor(verbosity=2, max_time_mins=2, population_size=50)
  tpot <- fit(tpot, train_wine.features, train_wine.classes)
  expect_true(BBmisc::isSubset(c("TPOTRRegressor"), class(tpot)))
  pred <- predict(tpot, test_wine.features)
  s = score(tpot, test_wine.features, test_wine.classes)
  expect_true(is.numeric(s))
})