Customer type discovery in hotel revenue management by Memetic algorithm

In case of using sales transaction and products availability data, discovering a preference list of customers (named customer types) is a challenging topic due to the lack of data. This process is inferred as demand estimation because by knowing the preference of each customer and available products in a time period in the future, we can estimate the demand of each product in case of arriving pre-known customer types. In this paper, we proposed an approach to find the frequency of each data point (solution) in a search space, helping us to apply a better local search in an evolutionary algorithm. Here, we proposed a memetic algorithm to discover customer types with the help of found frequency data. Afterward, we used five real hotel datasets to implement our memetic algorithm and finally compared this implementation with the two pre-proposed approaches, one evolutionary approach (genetic), and one mathematical approach (rank based). Evaluation proceeds in two phases in both of which, memetic shows a better performance than the other two approaches owing to using local search done with help of frequency data.