|
Activity Number:
|
376
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Business and Economic Statistics Section
|
| Abstract - #303913 |
|
Title:
|
Automatic Time Series Model Selection
|
|
Author(s):
|
Dongik Jang*+ and Hee-Seok Oh
|
|
Companies:
|
Seoul National University and Seoul National University
|
|
Address:
|
Gwanak 599, Gwanak-ro, Gwanak-gu, Seoul, International, 151-742, South Korea
|
|
Keywords:
|
Automatic model selection ; ARIMA models ; Exponential smoothing ; Threshold AR models ; GARCH models ; R
|
|
Abstract:
|
In this paper, we propose automatic procedures for selecting time series model from various univariate time series data. Handling model selection in an automatic way is important when model selection is applied in a `data mining' context to the large number of time series such as the thousands of signals to access a web sever in a specific time period. Previously, several methods have been proposed to select model selection of time series automatically. However, most of existing methods mainly focus on linear time series including exponential smoothing and ARIMA models. We describe automatic model selection algorithms that have been implemented in an {\tt autots} package for R. The key feature that distinguishes the proposed procedures from previous approaches is that the former can handle linear time series as well as nonlinear time series such as threshold AR models and GARCH models.
|
