Businesses may find it useful to model and predict how customers will choose among available products. A conditional logit model can be used for modeling choices between many alternatives. This flexible model allows for attributes to vary over individuals and alternatives. This model can be used to predict customers' willingness to pay for each available choice.

People often face hierarchical choices, such as products of varying quality, but are contrained by their resources. Given observed choices between three ranked alternatives, you can use Gibbs sampling with data augmentation to fit an ordered probit model. This Bayesian procedure generates a probability distribution for the model parameters. Businesses may find this model useful for predicting the likelihood of customers choosing low, medium, or premium grade products.

Business are often confronted with situations where one of two possibilities may occur, and may find it strategic to estimate the probability of each outcome. The Metropolis-Hastings algorithm, an advanced Markov chain Monte Carlo method that simulates draws from a Bayesian posterior distribution, is a powerful tool that can be applied for this purpose. Given observed data and a proposed model, the method generates a probability distribution for the model parameters. An example is provided that fits a logit model, which can be used to model binary choices. Business may find this model useful for its power in predicting the likelihood of each of two alternatives occuring.