Library: | times |
See also: | armalik armapq |
Quantlet: | armacls | |
Description: | estimates an autoregressive moving average process with zero mean by maximizing the conditional sum of squared residuals. |
Usage: | {y,wnv} = armacls(x,p,q{,maxiter,eps}) | |
Input: | ||
x | n x 1 vector, the process (time series) | |
p | scalar, the autoregression order | |
q | scalar, the moving average order | |
maxiter | (optional) scalar, maximal number of iterations; default: maxiter = 250 | |
eps | (optional) scalar, convergence tolerance of the function value; default: eps = 1e-7 | |
Output: | ||
y | list containing 1. (p+q) x 1 vector, the estimated parameters, 2. the number of iterations, and 3. a scalar indicating whether convergence occurred (=1) or not (=0) | |
wnv | scalar, the estimate of the white noise variance |
library("times") ; loads the quantlets from Times Library randomize(0) ; sets random seed x = genarma(0.7|0.1,0.3,normal(500)) ; Generates ARMA(2,1) with White Noise {y,wnv} = armacls(x,2,1,50) ; Estimation procedure for above ARMA(2,1) process y wnv randomize(7) {y,wnv} = armacls(x,2,1,50) ; Reestimation with different start values y wnv
Contents of y.minimum ; vector with estimated parameters [1,] 0.56378 ; AR(1) [2,] 0.22206 ; AR(2) [3,] 0.40891 ; MA(1) Contents of y.iter ; Number of iterations till convergence [1,] 32 Contents of y.converged ; Convergence achieved: YES [1,] 1 Contents of wnv [1,] 0.91065 Contents of y.minimum ; vector with estimated parameters [1,] 0.56627 ; AR(1) [2,] 0.22044 ; AR(2) [3,] 0.40804 ; MA(1) Contents of y.iter [1,] 44 ; Number of iterations till convergence Contents of y.converged [1,] 1 Contents of wnv [1,] 0.91065