sandbox/cvglmnet.R

In JustinMShea/ExpectedReturns: Reproduction of Investment Strategies presented in Antti Ilmanen's Expected Returns and Investing Amid Low Expected Returns

mlr_cvglmnet <- function(training_data,
                     y,
                     alpha = 0.5) {
  enet <- TaskRegr$new(id = "enet", backend = training_data, target = y)
  lrn_enet <- lrn("regr.cv_glmnet", alpha = alpha)
  lrn_enet$train(enet)
  return(lrn_enet)
}

test_cvglmnet <- function(training_data,
                          testing_data,
                          y,
                          alpha = 0.5) {
  x_test <- as.matrix(testing_data[, !..y])

  mlr_enet <- mlr_cvglmnet(training_data, y, alpha)
  mlr_enet_fitted <- mlr_enet$model
  mlr_enet_lambda_min <- mlr_enet_fitted$lambda.min

  mlr_enet_predict <- predict(mlr_enet_fitted, newx = x_test, s = mlr_enet_lambda_min)

  x_train <- as.matrix(training_data[, !..y])
  y_train <- training_data[[y]]

  glm_enet <- cv.glmnet(x_train, y_train, alpha = alpha)
  glm_enet_lambda_min <- glm_enet$lambda.min

  glm_predict <- predict(glm_enet, newx = x_test, s = glm_enet_lambda_min)

  if (identical(mlr_enet_predict, glm_predict)) {
    print("cv.glmnet test passed.")
  } else {
    stop("Warning: cv.glmnet test failed.")
  }
}