Advances and Applications in Statistics
Volume 4, Issue 2, Pages 233 - 251
(August 2004)
|
|
TIME SERIES LAG SELECTION USING GENETIC ALGORITHM
Ashok K. Nag (India) and Amit Mitra (India)
|
Abstract: In
this paper, we consider the fundamental
problem of model selection in the time series
framework. A new method based on artificial
intelligence techniques is proposed. The
method uses genetic optimization search and
artificial neural network model building
techniques. The main advantage of using the
proposed method lies in the fact that it does
not make any theoretical assumptions on the
data generating process allowing the data to
speak for themselves. Extensive simulation
study is performed to investigate the
usefulness of the proposed method through
simulated auto regressive (AR) time series
models. We also compare the performance of the
proposed procedure with commonly used
information theoretic methods. Results from
the simulations indicate that the proposed
procedure performs significantly better than
the existing methods for model selection based
on information theoretic criterion. |
Keywords and phrases: Akaike information criterion, artificial neural network, genetic algorithm, information theoretic criterion, lag selection, minimum description length, Monte Carlo simulation, Schwarz Bayesian criterion. |
|
Number of Downloads: 359 | Number of Views: 1313 |
|