Title: Investigating Q-learning approach by using reinforcement learning to decide dynamic pricing for multiple products

Authors: Fakhraddin Maroofi

Addresses: Department of Business Administration, University of Kurdistan, Sanandaj, Iran

Abstract: This article emphasises the benefits of using mutually dynamic pricing, as opposed individual pricing of product or services. By using mutually beneficial rating, the algorithm is able to use the information of various product or services to enhance the profit received from rating all the items in a consistent manner. This enables for quicker learning once the demand for the various product or services is powerfully connected. However, the range of mutually beneficial product will increase the speed of convergence decreases exponentially. Because the range of mutually beneficial product becomes large, the decision maker could take into account grouping product if they follow an equal demand pattern, or put together rating extremely related mutually beneficial product. Moreover, we analyse to behave the Q-learning with eligibility trace algorithm under different conditions without any explicit knowledge of client buying behaviour.

Keywords: Q-learning approach; reinforcement learning; service management; simulation; Iran.

DOI: 10.1504/IJBIS.2019.099528

International Journal of Business Information Systems, 2019 Vol.31 No.1, pp.86 - 105

Received: 10 Oct 2016
Accepted: 26 Aug 2017
Published online: 08 May 2019 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article