Analysis of green manure decomposition parameters in northeast Brazil using association rule networks

Abstract The food sector is one of the most critical areas of the economy, and consumers are seeking safer, more readily available, more affordable, and better quality food. Therefore, organic agriculture has become a possible approach for optimizing the characteristics of processed foods. Vegetables have essential uses as green manure, but the greatest difficulty encountered when using these species is related to the time required for their residues to decompose. Computational intelligence and data mining techniques are used widely in agricultural studies and for process improvement purposes. Association rules are used as data mining techniques to identify patterns in large databases, where the patterns identified are analyzed and transformed into a graph to facilitate further analyses. In this study, we employed the patterns identified by association rule networks (ARNs) to determine directly the key parameters related to the half-life decomposition times of legumes used as green manure in northeast Brazil. We validated this method by comparing the results obtained by the ARNs and a decision tree. The proposed approach obtained promising results, thereby demonstrating its capacity to describe a set of objective items and facilitate the development of more integrated hypotheses. Thus, it was possible to optimize the discovery of the key parameters related to green fertilizers in order to identify the best species according to its culture requirements and to enhance productivity.