roll_predict_l2relax: Rolling window forecast

In zhan-gao/LasForecast: Time series linear predictive regression

Rolling window forecast

Description

The function reads data and make forecasts based on linear predictive regression with diverse methods. It incorporates both short-horizon and long-horizon forecasting.

Usage

roll_predict_l2relax(
  x,
  y,
  roll_window,
  h = 1,
  k_max = 4,
  m = 5,
  ntau = 100,
  tau_min_ratio = 0.01,
  train_method = "oos",
  solver = "CVXR",
  tol = 1e-07,
  verb = TRUE,
  csr = TRUE
)

Arguments

x	Full sample predictor
y	Full sample forecast target
roll_window	Length of the rolling window
h	forecast horizon
m	number of folds
ntau	number of tau values
train_method	parameter tuning method for L2relax "cv_random", "cv" or "oos"
solver	"Rmosek" or "CVXR"
tol	tolerance for the solver
verb	boolean to control whether print information on screen
csr	boolean to opt out for the csr
tau.min.ratio	ratio of the minimum tau in tau.seq over the maximum (which is the smallest tau such that equal-weight solves the forecast combination optimization.)

zhan-gao/LasForecast documentation built on Sept. 18, 2024, 9:40 p.m.