Customer type discovery in hotel revenue management: a data mining approach

Abstract

Demand estimation is a fundamental component of revenue management systems. The demand for a product can be ascertained from the customers who purchase it. Identifying customer types in this context is a challenging endeavor, recently resolved using meta-heuristic and mathematical techniques. Meta-heuristics leverage the scarcity of data in the search space, commencing with random samples and employing the fitness function as a guide during operations. Our proposed approach generates the search space by incorporating supplementary data to identify valuable customer types. We employ a new period table with additional data to achieve this objective. Subsequently, we reduce the search space through data mining's clustering method and ultimately employ a greedy algorithm and fitness function to identify valuable customer types and construct our solution. To validate our approach, we compare our solution and the most recent research in this field, including genetic, memetic, and mathematical approaches. Compared to memetic methods, our results indicate that our solution has a smaller length, with a maximum reduction of 34%, and exhibits improvement in log value, with a maximum of 7%.