l1_p: l1_p

In Monster-Moon/l1pm: The l1 Penalization Method

l1_p

Description

Estimates non crossing quantile regression with a neural network.

Usage

l1_p(
  X,
  y,
  test_X,
  valid_X,
  tau,
  hidden_dim1,
  hidden_dim2,
  learning_rate,
  max_deep_iter,
  lambda_obj,
  penalty = 0
)

Arguments

X	train predictor data
y	train response data
test_X	test predictor data
valid_X	validation predictor data
tau	target quantiles
hidden_dim1	the number of nodes in the first hidden layer
hidden_dim2	the number of nodes in the second hidden layer
learning_rate	learning rate in the optimization process
max_deep_iter	the number of iterations
lambda_obj	the value of tuning parameter in the l1 penalization method
penalty	the value of tuning parameter for ridge penalty on weights

Value

y_predicted, y_test_predicted, y_valid_predited : predicted quantile based on train, test, and validation data, respectively

Monster-Moon/l1pm documentation built on July 20, 2024, 6:55 p.m.