Despite the potential of association rules to extract knowledge in datasets, the selection of valuable rules has been the specialists’ big challenge because of the great number of rules generated by association rules’ algorithms. Unfortunately, for Species Distribution Modeling, researchers and analysts alike has been discourage regarding the efficacy and relative merits to the studies due to the conflicting of results. In this case study, we integrate geoprocessing techniques with association rules to analyze the potential distribution of the charcoal tree species of Amazon, Tachigali Subvelutina. This integration allowed the exploration of a new method to select valuable association rules based not only by frequent item sets, like confidence, support and lift. But also considering the geospatial distribution of the species occurrences. The application of one more dimensional to the measures of the frequent item sets provides a more effective understanding and adequate amount of relevance to represent significant association rules. Further, this new approach can support experts to: identify effective rules, i.e., rules that can be effective used with regression techniques to predict species in geospatial environments; identify variables that play a decisive role to discover the occurrence of species; and evaluate the effectivity of the datasets used for the study to answer the specified scientific questions.
CITATION STYLE
Prata, D. N., Rocha, M. L., Ferreira, L. O., & Nogueira, R. (2019). Geospatial Dimension in Association Rule Mining: The Case Study of the Amazon Charcoal Tree. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11943 LNCS, pp. 244–258). Springer. https://doi.org/10.1007/978-3-030-37599-7_21
Mendeley helps you to discover research relevant for your work.