In the context of customer satisfaction and the delivery of accurate results, recommendation systems have emerged as a significant tool. These systems leverage advanced data analysis methodologies, machine learning algorithms, and neural networks to deliver personalised recommendations optimised for individual users. The underlying principle is to predict user preferences by analysing their historical behaviours and establishing similarities with other users in the system. The objective of this study is to evaluate the neighbourhood provided for a session-based recommendation system that has been generated by a density-based clustering algorithm. According to the articles [4], the cluster-based approach offers greater accuracy and diversity in recommendation lists than kNN. The experiments described in this article are related to a density-based clustering. The final results were evaluated in terms of accuracy, coverage and popularity.