A comparative study of information criteria for model selection

被引:16
作者
Nakamura, Tomomichi
Judd, Kevin
Mees, Alistair I.
Small, Michael
机构
[1] Univ Western Australia, Ctr Appl Dynam & Optimizat, Sch Math & Stat, Nedlands, WA 6009, Australia
[2] Predict Co, Santa Fe, NM 87501 USA
[3] Hong Kong Polytech Univ, Dept Elect & Informat Engn, Kowloon, Hong Kong, Peoples R China
来源
INTERNATIONAL JOURNAL OF BIFURCATION AND CHAOS | 2006年 / 16卷 / 08期
关键词
information criteria; fitting errors; model selection; the least squares method;
D O I
10.1142/S0218127406015982
中图分类号
O1 [数学];
学科分类号
0701 ; 070101 ;
摘要
To build good models, we need to know the appropriate model size. To handle this problem, a variety of information criteria have already been proposed, each with a different background. In this paper, we consider the problem of model selection and investigate the performance of a number of proposed information criteria and whether the assumption to obtain the formulae that fitting errors are normally distributed hold or not in some conditions (different data points and noise levels). The results show that although the application of information criteria prevents over-fitting and under-fitting in most cases, there are cases where we cannot avoid even involving many data points and low noise levels in ideal situations. The results also show that the distribution of the fitting errors is not always normally distributed, although the observational noise is Gaussian, which contradicts an assumption of the information criteria.
引用
收藏
页码:2153 / 2175
页数:23
相关论文
共 33 条
[1]   NEW LOOK AT STATISTICAL-MODEL IDENTIFICATION [J].
AKAIKE, H .
IEEE TRANSACTIONS ON AUTOMATIC CONTROL, 1974, AC19 (06) :716-723
[2]   The minimum description length principle in coding and modeling [J].
Barron, A ;
Rissanen, J ;
Yu, B .
IEEE TRANSACTIONS ON INFORMATION THEORY, 1998, 44 (06) :2743-2760
[3]  
BAXTER RA, 1994, 207 MON U DEP COMP S
[4]   NONLINEAR PREDICTION OF CHAOTIC TIME-SERIES [J].
CASDAGLI, M .
PHYSICA D, 1989, 35 (03) :335-356
[5]   ORTHOGONAL LEAST-SQUARES LEARNING ALGORITHM FOR RADIAL BASIS FUNCTION NETWORKS [J].
CHEN, S ;
COWAN, CFN ;
GRANT, PM .
IEEE TRANSACTIONS ON NEURAL NETWORKS, 1991, 2 (02) :302-309
[6]   PREDICTING CHAOTIC TIME-SERIES [J].
FARMER, JD ;
SIDOROWICH, JJ .
PHYSICAL REVIEW LETTERS, 1987, 59 (08) :845-848
[7]   2-DIMENSIONAL MAPPING WITH A STRANGE ATTRACTOR [J].
HENON, M .
COMMUNICATIONS IN MATHEMATICAL PHYSICS, 1976, 50 (01) :69-77
[8]   Embedding as a modeling problem [J].
Judd, K ;
Mees, A .
PHYSICA D, 1998, 120 (3-4) :273-286
[9]  
Judd K, 2003, CHAOS AND ITS RECONSTRUCTION, P179
[10]   Modeling chaotic motions of a string from experimental data [J].
Judd, K ;
Mees, A .
PHYSICA D, 1996, 92 (3-4) :221-236