Robust Multivariate Exponential Smoothing

Share:

Description

Robust Multivariate Exponential Smoothing

Usage

1
2
3
robustMultExpSmoothing(R, smoothMat = NA, startup_period = 10,
  training_period = 60, seed = 9999, trials = 50, method = "L-BFGS-B",
  lambda = 0.2)

Arguments

R

data

smoothMat

Optimal smoothing matrix. If missing it is estimated. The procedure maybe very slow for high-dimensional data. Also, the objective function being very noisy, optimization across multiple runs may lead to different smoothing matrices. #'

startup_period

length of samples required to calculate initial values

training_period

length of samples required to calculate forecast errors for evalualating the objective if smoothing matrix is estimated

seed

random seed to replicate the starting values for optimization

trials

number of strarting values to try for any optimization. Large number of trials for high dimensions can be time consuming

method

optimization method to use to evaluate an estimate of smoothing matrix. Default is L-BFGS-B

lambda

known constant as described in the paper. Defaults to 0.2

Details

Calculate Robust estimate of covariance matrix while also smoothing and cleaning the data using the procedure described in (Croux, Gelper, and Mahieu, 2010)

Author(s)

Rohit Arora