Hidden Markov models and their applications to customer relationship management

Hidden Markov models (HMMs) are widely used in science, engineering and many other areas. In an HMM, there are two types of states: hidden states and observable states. Here we propose an HMM via the framework of a Markov chain model. Simple estimation methods for the transition probabilities among the hidden states are discussed. The estimation methods are better than the traditional EM algorithm in both the quality of estimation and the computational complexity. We then apply the model to classify the customers of a computer service company which is an important task in the customer relationship management. Numerical examples are given to illustrate the usefulness of the model by using a real-world data set.