Effect of aggregation functions on the habitat preference modelling using a genetic Takagi-Sugeno fuzzy system

Uncertainties originating from the behaviour of target species and modelling approaches affect predictive accuracy and information retrieved, which can thus influence the applicability and reliability of a model. This paper aimed to assess the effects of aggregation functions for computing composite habitat preference on the prediction of species distributions and habitat preference evaluation using a 0-order genetic Takagi-Sugeno fuzzy model. The effects were evaluated based on the predictive accuracy and habitat preference information. In order to reduce the data uncertainty, artificial data were generated using hypothetical habitat preference curves (HPCs) under different assumptions on the interaction between habitat variables and habitat preference of an artificial fish. In total, twelve data sets were generated, from which forty-eight fuzzy habitat preference models (FHPMs) with different aggregation functions were developed. As a result, the FHPMs produced similar HPCs across the different data sets, while slight differences were found between the FHPMs with different aggregation functions. Although none of the models could represent hypothetical habitat preference, the product-type aggregation function showed relatively higher performance for both accuracy and HPCs.