Sampling Experiments and the Quandt Statistic

J M Freeman

Sampling Experiments and the Quandt Statistic

J M Freeman

MSM Decision Sciences

Research output: Contribution to journal › Article › peer-review

Abstract

The distribution of the Quandt likelihood ratio, lambda t, for a two-phase regression, has yet to be determined. In particular, it is known that -2log (lambda t) is not distributed as chisquare (Quandt, 1960), for unknown switch point. In this paper we describe sampling experiments which suggest that - log (lambda t) has a Pearson Type III distribution. The parameters appear to depend not only on the values of the x vector (Feder, 1969) but also its dimension, k.

Original language	English
Pages (from-to)	1879-1887
Journal	Communications in Statistics.
Volume	12
Issue number	16
Publication status	Published - 1983

Keywords

regression, segmented, likelihood

Cite this

@article{79b28e7a6cf848d1b156ee082c6b8334,

title = "Sampling Experiments and the Quandt Statistic",

abstract = "The distribution of the Quandt likelihood ratio, lambda t, for a two-phase regression, has yet to be determined. In particular, it is known that -2log (lambda t) is not distributed as chisquare (Quandt, 1960), for unknown switch point. In this paper we describe sampling experiments which suggest that - log (lambda t) has a Pearson Type III distribution. The parameters appear to depend not only on the values of the x vector (Feder, 1969) but also its dimension, k.",

keywords = "regression, segmented, likelihood",

author = "Freeman, {J M}",

year = "1983",

language = "English",

volume = "12",

pages = "1879--1887",

journal = "Communications in Statistics.",

issn = "1532-415X",

publisher = "Taylor & Francis",

number = "16",

}

TY - JOUR

T1 - Sampling Experiments and the Quandt Statistic

AU - Freeman, J M

PY - 1983

Y1 - 1983

N2 - The distribution of the Quandt likelihood ratio, lambda t, for a two-phase regression, has yet to be determined. In particular, it is known that -2log (lambda t) is not distributed as chisquare (Quandt, 1960), for unknown switch point. In this paper we describe sampling experiments which suggest that - log (lambda t) has a Pearson Type III distribution. The parameters appear to depend not only on the values of the x vector (Feder, 1969) but also its dimension, k.

AB - The distribution of the Quandt likelihood ratio, lambda t, for a two-phase regression, has yet to be determined. In particular, it is known that -2log (lambda t) is not distributed as chisquare (Quandt, 1960), for unknown switch point. In this paper we describe sampling experiments which suggest that - log (lambda t) has a Pearson Type III distribution. The parameters appear to depend not only on the values of the x vector (Feder, 1969) but also its dimension, k.

KW - regression, segmented, likelihood

M3 - Article

SN - 1532-415X

VL - 12

SP - 1879

EP - 1887

JO - Communications in Statistics.

JF - Communications in Statistics.

IS - 16

ER -

Sampling Experiments and the Quandt Statistic

Abstract

Keywords

Fingerprint

Cite this