test-ml.model.step.lm.R

#
# (C) Copyright IBM Corp. 2017
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#

context("Testing r4ml.step.lm\n")

test_that("r4ml.step.lm direct", {
  data("iris")

  r4ml_iris <- as.r4ml.matrix(iris[, -5])
  step_lm <- r4ml.step.lm(Sepal_Length ~ ., data = r4ml_iris)
  coef(step_lm)
  stats(step_lm)
})

test_that("r4ml.step.lm with intercept", {
  data("iris")

  r4ml_iris <- as.r4ml.matrix(iris[, -5])
  step_lm <- r4ml.step.lm(Sepal_Length ~ ., data = r4ml_iris, intercept = TRUE)
  coef(step_lm)
  stats(step_lm)
})

test_that("r4ml.step.lm predict", {
  df <- iris
  df$Species <- (as.numeric(df$Species))
  iris_df <- as.r4ml.frame(df)
  iris_mat <- as.r4ml.matrix(iris_df)
  ml.coltypes(iris_mat) <- c("scale", "scale", "scale", "scale", "nominal")
  s <- r4ml.sample(iris_mat, perc=c(0.2,0.8))
  test <- s[[1]]
  train <- s[[2]]
  y_test <- as.r4ml.matrix(test[, 1])
  y_test <- SparkR:::as.data.frame(y_test)
  test <- as.r4ml.matrix(test[, c(2:5)])
  iris_step_lm <- r4ml.step.lm(Sepal_Length~ . , data = train)
  output <- predict(iris_step_lm, test)
  expect_true( base::mean(sapply(SparkR::as.data.frame(output[[1]])-y_test, abs)) < 5)
})

test_that("r4ml.step.lm predict_scoring", {
  df <- iris
  df$Species <- (as.numeric(df$Species))
  iris_df <- as.r4ml.frame(df)
  iris_mat <- as.r4ml.matrix(iris_df)
  ml.coltypes(iris_mat) <- c("scale", "scale", "scale", "scale", "nominal")
  s <- r4ml.sample(iris_mat, perc=c(0.2,0.8))
  test <- s[[1]]
  train <- s[[2]]
  y_test <- as.r4ml.matrix(test[,1])
  y_test <- SparkR:::as.data.frame(y_test)
  test <- as.r4ml.matrix(test[, c(2:5)])
  iris_step_lm <- r4ml.step.lm(Sepal_Length ~ ., data=train)
  output <- predict(iris_step_lm, test)
  expect_true( mean(sapply(SparkR::as.data.frame(output[[1]])-y_test, abs)) < 5)
})

# TODO add the shiftAndScale test in future when intercept=2 is added

test_that("r4ml.step.lm coefficients without intercept", {
  df <- mtcars
  r.slm <- step(lm(mpg ~ -1, data = df),
                mpg ~ cyl + disp + hp + drat + wt + qsec + vs + am + gear + carb,
                direction = "forward")
  
  df.r4ml <- as.r4ml.matrix(as.r4ml.frame(df))
  r4ml.slm <- r4ml.step.lm(mpg ~ ., data = df.r4ml, intercept = FALSE)
  
  # coefficients from step() and r4ml.step.lm() are not in the same order
  # So, we'll sort the coefficients by name for both models before comparison
  snames <- sort(names(coef(r.slm)))

  betas1 <- coef(r4ml.slm)[snames,]
  betas2 <- coef(r.slm)[snames]
  names(betas2) <- NULL

  ## expect_equal(betas1, betas2) # TODO debug later
})

test_that("r4ml.step.lm coefficients with intercept", {
  df <- mtcars
  r.slm <- step(lm(mpg ~ 1, data = df),
                mpg ~ cyl + disp + hp + drat + wt + qsec + vs + am + gear + carb,
                direction = "forward")
  
  df.r4ml <- as.r4ml.matrix(as.r4ml.frame(df))
  r4ml.slm <- r4ml.step.lm(mpg ~ ., data = df.r4ml, intercept = TRUE)
  
  # coefficients from step() and r4ml.step.lm() are not in the same order
  # So, we'll sort the coefficients by name for both models before comparison
  snames <- sort(names(coef(r.slm)))
  
  betas1 <- coef(r4ml.slm)[snames,]
  betas2 <- coef(r.slm)[snames]
  names(betas2) <- NULL
  
  ## expect_equal(betas1, betas2) # TODO debug later
})