Crevettic: uma plataforma de business intelligence para gestão de carcinicultura

Abstract

Farm shrimp production is an important sector in the economy of Northeastern Brazil, mainly in the states of Ceará and Rio Grande do Norte, which together produce 66% of annual national production, about 27.2 thousand tons in total. However, the emergence of new diseases has forced producers to carry out more effective monitoring in relation to the management and health of shrimps. Several information is collected in order to previously identify signs that may cause losses of production, such as slow development and increased mortality rate. However, many farms still control these data inefficiently, using paper or spreadsheets that grows considerably, making them difficult to analyze for humans. Thus, decision-making on farms ends up being based on trial and error, by testing various parameters without scientific basis. In this context, in order to fill out this sector demand, this work proposes the Crevettic platform, a solution focused on knowledge management and business intelligence on shrimp farms, presenting indicators to producers through data analysis using data mining and machine learning. As a proof of concept, two prediction algorithms were trained. First one aimed at predicting the average weight of the shrimp in the tank, with a determination coefficient of 94%. Second one aimed at predicting the ideal necessary ration in the tank, reaching 98% in the coefficient of determination. These information is presented on the platform’s dashboard. The platform was developed using the microservice architecture (divided into modules) and is currently being used by seven peaceful farms in various states in the country. The Crevettic platform proposal went through several pre-acceleration and acceleration notices, evolving into a commercial product, which later became a startup