A Hidden Markov Model of Customer Relationship Dynamics

Dynamic travel mode searching and switching analysis considering hidden model preference and behavioral decision processes

“…the hidden-state transition). It has been applied successfully to, e.g., speech recognition, biological sequences analysis, and many others (Netzer et al 2008;Scott 2002). The objective of this paper is to develop an individuallevel dynamic model that explicitly parameterizes the processes that travelers use to search and identify their alternative modes.…”

“…(3). According to the Ergodic Theory, these properties guarantee the existence and uniqueness of the stationary distribution (Netzer et al, 2008;Wahba and Shalaby 2014), which ensures a unique initial-state distribution obtained from solving Eq. (4).…”

Modeling strategic group dynamics: A hidden Markov approach

“…We adopt a method proposed by Chib (1995) to calculate the marginal density for our hidden Markov model from the output of the MCMC sampler above. Second, we use the Markov Switching Criterion (MSC) which has been developed for HMMs by Smith et al (2006) and adapted by Netzer et al (2008) for Bayesian HMMs. Third, we calculate the posterior probability of each model with k=1,…,K components directy from the MCMC output using the method outlined by Gamerman and Lopes (2006).…”

“…For that purpose, we propose the use of Hidden Markov Models (HMM) (Du and Kamakura 2006;Frühwirth-Schnatter 2006;Netzer et al 2008) that are able to simultaneously idenfity latent states (i.e., strategic groups), and explitly model the transisition probabilites across the identified strategic groups to account for such potential time dependencies. Thus, whereas exisiting methods identify market structures from available strategic variables independently for each time period, the proposed method accounts for time dependence through a HMM that enables us to identifty how strategic groups evolve over time.…”

“…Thus, with this hidden Markov approach, our objective is to explictly model how strategy and membership of strategic groups vary over time (Mascarenhas 1989). The (Bayesian) HMM framework has been found to be a powerfull and flexible approach to identify time dependencies and (latent) group forming simultaneously, and has been recently applied in various areas of marketing by Du and Kamakura (2006) who use HMMs to identify stages of famlily lifecylces, Netzer et al (2008) who model the dynamics of customer relationships, Moon et al (2007) who use HMMs to estimate the effect of sales promotions, and van der Lans et al (2008) and Liechty et al (2003) to model eye-movement data. Our study demonstrates how HMMs may be used to study competition for strategic marketing planning.…”