手法を比較

	Scanner Panel Analysis	Hierarchical Bayes Choice Model
分野	マーケティング	マーケティング
系統	Regression model	Regression model
提唱年	1983	2005
提唱者	Peter M. Guadagni & John D. C. Little	Peter E. Rossi, Greg M. Allenby & Robert McCulloch
種類	Disaggregate multinomial-logit brand-choice model	Hierarchical Bayesian random-coefficients discrete-choice model
原典	Guadagni, P. M., & Little, J. D. C. (1983). A Logit Model of Brand Choice Calibrated on Scanner Data. Marketing Science, 2(3), 203-238.	Rossi, P. E., Allenby, G. M., & McCulloch, R. (2005). Bayesian Statistics and Marketing. John Wiley & Sons. ISBN: 9780470863671
別名	Scanner Panel Logit, Guadagni-Little Model, Household Panel Choice Model, Loyalty-Variable Logit	HB Choice Model, Bayesian Random-Coefficients Logit, Hierarchical Bayesian Conjoint, Individual-Level Partworth Model
関連	3	3
概要	Scanner panel analysis models individual households' brand choices using the purchase histories captured by UPC scanner panels, in which the same households are tracked occasion by occasion with the brand chosen and the prices and promotions they faced. The defining method is Guadagni and Little's 1983 multinomial logit of brand choice, the first model to put scanner panel data to serious analytical use. Its signal innovation is the loyalty variable: an exponentially smoothed measure of each household's past brand purchases that enters the utility function and captures persistent brand preference and state dependence. Alongside loyalty, the model includes price, promotion, and brand intercepts, and yields the probability that a household buys each brand on a given occasion. From the fitted model one recovers price and promotion elasticities at the individual level and can simulate how marketing actions shift choices. The framework launched the modern era of disaggregate choice modeling and remains the reference point for scanner-based brand-choice analysis.	Hierarchical Bayes (HB) choice models estimate a separate set of preference weights — partworths — for every individual respondent, while borrowing strength across respondents through a shared population distribution. The model has two levels: at the lower level each person's choices follow a logit driven by their own coefficients, and at the upper level those individual coefficients are treated as draws from a common multivariate distribution whose mean and covariance are themselves estimated. Inference is Bayesian and proceeds by Markov chain Monte Carlo — typically Gibbs sampling with Metropolis steps — which yields a full posterior for each respondent's partworths rather than a single point estimate. The approach, codified by Rossi, Allenby, and McCulloch, solved a long-standing problem in choice modeling: how to recover genuine individual-level heterogeneity from the sparse data each person provides. Sparse individual estimates are stabilized by shrinkage toward the population mean, giving reliable person-level coefficients usable for segmentation, targeting, and realistic market simulation. HB is now the default estimator for conjoint and scanner-based choice analysis.