A Bayesian methodology for simultaneously detecting and estimating regime change points and variable selection in multiple regression models for marketing research

Duncan K.H. Fong, Wayne S. DeSarbo

Research output: Contribution to journalArticle

10 Citations (Scopus)

Abstract

We present a Bayesian change point multiple regression methodology which simultaneously estimates the location of change points/regimes, the corresponding subset of independent variables per regime, as well as the associated regimes' regression parameters. Unlike existing switching multiple regression models, our method does not require the presence of all independent variables in each regime to detect change points. This allows us to relax the minimum size constraint on each regime as fewer observations are needed to estimate the unknown regression coefficients. Thus our method provides a means to search for small regimes where only a few independent variables are significant. Note that accuracy of change points can drastically affect the identified models within each regime. As the number of change points in the data is typically unknown, we have provided a probability based model selection heuristic to determine its value. Both synthetic and real data sets are utilized to demonstrate that our procedure can yield better fitted models over aggregate OLS regression models and traditional MLE based regime switching models. Furthermore, an actual prescription drug data application involving a promotion response model is used to gainfully illustrate the methodology.

Original languageEnglish (US)
Pages (from-to)427-453
Number of pages27
JournalQuantitative Marketing and Economics
Volume5
Issue number4
DOIs
StatePublished - Dec 1 2007

Fingerprint

Change point
Multiple regression
Regime change
Marketing research
Regression model
Variable selection
Methodology
Coefficients
Prescription drugs
Heuristics
Regime-switching model
Model selection

All Science Journal Classification (ASJC) codes

  • Economics, Econometrics and Finance (miscellaneous)
  • Marketing

Cite this

@article{bd41ba83b1e5426cbaf5dd4865f1d47a,
title = "A Bayesian methodology for simultaneously detecting and estimating regime change points and variable selection in multiple regression models for marketing research",
abstract = "We present a Bayesian change point multiple regression methodology which simultaneously estimates the location of change points/regimes, the corresponding subset of independent variables per regime, as well as the associated regimes' regression parameters. Unlike existing switching multiple regression models, our method does not require the presence of all independent variables in each regime to detect change points. This allows us to relax the minimum size constraint on each regime as fewer observations are needed to estimate the unknown regression coefficients. Thus our method provides a means to search for small regimes where only a few independent variables are significant. Note that accuracy of change points can drastically affect the identified models within each regime. As the number of change points in the data is typically unknown, we have provided a probability based model selection heuristic to determine its value. Both synthetic and real data sets are utilized to demonstrate that our procedure can yield better fitted models over aggregate OLS regression models and traditional MLE based regime switching models. Furthermore, an actual prescription drug data application involving a promotion response model is used to gainfully illustrate the methodology.",
author = "Fong, {Duncan K.H.} and DeSarbo, {Wayne S.}",
year = "2007",
month = "12",
day = "1",
doi = "10.1007/s11129-007-9030-8",
language = "English (US)",
volume = "5",
pages = "427--453",
journal = "Quantitative Marketing and Economics",
issn = "1570-7156",
publisher = "Kluwer Academic Publishers",
number = "4",

}

TY - JOUR

T1 - A Bayesian methodology for simultaneously detecting and estimating regime change points and variable selection in multiple regression models for marketing research

AU - Fong, Duncan K.H.

AU - DeSarbo, Wayne S.

PY - 2007/12/1

Y1 - 2007/12/1

N2 - We present a Bayesian change point multiple regression methodology which simultaneously estimates the location of change points/regimes, the corresponding subset of independent variables per regime, as well as the associated regimes' regression parameters. Unlike existing switching multiple regression models, our method does not require the presence of all independent variables in each regime to detect change points. This allows us to relax the minimum size constraint on each regime as fewer observations are needed to estimate the unknown regression coefficients. Thus our method provides a means to search for small regimes where only a few independent variables are significant. Note that accuracy of change points can drastically affect the identified models within each regime. As the number of change points in the data is typically unknown, we have provided a probability based model selection heuristic to determine its value. Both synthetic and real data sets are utilized to demonstrate that our procedure can yield better fitted models over aggregate OLS regression models and traditional MLE based regime switching models. Furthermore, an actual prescription drug data application involving a promotion response model is used to gainfully illustrate the methodology.

AB - We present a Bayesian change point multiple regression methodology which simultaneously estimates the location of change points/regimes, the corresponding subset of independent variables per regime, as well as the associated regimes' regression parameters. Unlike existing switching multiple regression models, our method does not require the presence of all independent variables in each regime to detect change points. This allows us to relax the minimum size constraint on each regime as fewer observations are needed to estimate the unknown regression coefficients. Thus our method provides a means to search for small regimes where only a few independent variables are significant. Note that accuracy of change points can drastically affect the identified models within each regime. As the number of change points in the data is typically unknown, we have provided a probability based model selection heuristic to determine its value. Both synthetic and real data sets are utilized to demonstrate that our procedure can yield better fitted models over aggregate OLS regression models and traditional MLE based regime switching models. Furthermore, an actual prescription drug data application involving a promotion response model is used to gainfully illustrate the methodology.

UR - http://www.scopus.com/inward/record.url?scp=34848915733&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=34848915733&partnerID=8YFLogxK

U2 - 10.1007/s11129-007-9030-8

DO - 10.1007/s11129-007-9030-8

M3 - Article

AN - SCOPUS:34848915733

VL - 5

SP - 427

EP - 453

JO - Quantitative Marketing and Economics

JF - Quantitative Marketing and Economics

SN - 1570-7156

IS - 4

ER -