Abstract
The extraction of emerging patterns is a descriptive data mining technique using supervised learning. This means that emerging patterns describe knowledge or behaviour in data using labelled data. This paper proposes an exhaustive study of an evolutionary algorithm in order to extract fuzzy emerging patterns. Specifically, the proposal is a mono-objective evolutionary fuzzy system based on an iterative rule learning approach. This one shows a good behaviour and benefits with respect to other relevant algorithms within the literature in a complete experimental study supported by statistical tests.
The algorithm is also applied to a case study of an e-commerce retail business based on extra virgin olive oil: OrOliveSur. Due to the main properties of the emerging patterns, the study shows the interest of the knowledge extracted for this type of algorithm in order to compare statistics between different periods of time for the e-commerce website.
Experimental study
The following table presents the results obtained by the different algorithms. It must be noted that the EvAEFR is executed with 3 and 5 linguistic labels where the best results between both executions (3 or 5) for each dataset are highlighted. In fact, the values of these cells are presented in the contribution. It is important to remark that the best results are considered with respect to the GrothRate values obtained.