Quantum Approach for Market Basket Analysis on Qudit Systems

Abstract Quantum computing consists of a set of techniques based on the principles of quantum mechanics. Historically, quantum computing has been built around qubit systems featuring two discrete levels. However, qudit systems involving multiple levels appear to provide a promising solutions to certain difficulties related to noise and information capacity in quantum computing. This study investigate these advantages by extending the quantum approaches of market basket analysis (MBA) on qudit systems. Furthermore, we propose a novel method utilising an adaptive comparator that requires fewer quantum resources in assessing the minimum support threshold, made possible by qudit systems. The proposed quantum MBA method was simulated on an 8-level system for a database containing 9 items and 64 transactions. The experiment results show that the proposed method can perform MBA on the same database using 78,125% less quantum resources compared to existing methods. All simulations conducted in this study utilize the Cirq framework.