r_main_parallel_Matrix: Fit Parallelized QRADMM
In Scott-Coggeshall/bigrq: Parallelized Quantile Regression with ADMM

Description Usage Arguments

View source: R/main.R

r_main_parallel_Matrix is the model-fitting function for the parallelized QRADMM algorithm implemented entirely in R.

r_main_parallel_Matrix(
  X,
  y,
  M,
  intercept,
  max_iter = 500,
  min_iter = 10,
  n_workers,
  lambda,
  tau,
  rho,
  alpha,
  penalty = "lasso",
  penalized,
  inverses_write_path = NULL,
  inverses_read_path = NULL,
  abstol = 1e-07,
  reltol = 1e-04
)

`M`	an integer specifying the number of data blocks.
`intercept`	a logical indicating whether an intercept should be included in the model.
`max_iter`	an integer, the maximum number of iterations to run.
`min_iter`	an integer, the minimum number of iterations to run.
`n_workers`	an integer, the number of worker processes to create.
`lambda`	a numeric scalar or vector of non-zero values, the values for the tuning parameter.
`tau`	a numeric scalar between 0 and 1, the conditional quantile.
`rho`	a positive numeric scalar.
`alpha`	a positive numeric scalar.
`penalty`	a character string, the penalty to apply.
`penalized`	a logical vector of the same length as the number of model parameters, TRUE values indicate that the corresponding parameter should be penalized.
`inverses_write_path`	a character string, the location to save the calculated inverses to.
`inverses_read_path`	a character string, the location to read the calculated inverses from.
`reltol`
`dat`	a numeric matrix where the first column is the vector of responses and the rest of the columns are the design matrix for the model.