Business Intelligence for the detection of anomalies in records of fueling

Ano: 2018 | Volume: 3 | Número: 3

Autores: V. A. G. Aquize, M. M. Santos Filho

Palavras-chave: Pentaho BI; Machine Learning; Anomaly Detection; Decision Making; Data Warehouse

Resumo Inglês:

At present, in any organization it is necessary to make decisions, very strategic meetings to achieve a satisfactory development. It is the case of the National Hydrocarbon Agency of Bolivia (ANH), that due the smuggling of fuel, implemented the RFID technology in order to register the records of fueling of all fleet vehicular. From this, a model of Anomaly Detection in records of fueling was proposed trough machine learning techniques. Nevertheless, the huge volume of information about anomalies scores in a local and global level needs to analyzed and take decisions (e.g. strict control in some risk zones according variables analyzed). The collection and analysis of this information, given its heterogeneous character and its volume, usually become a problem for the government institution and this is where Business Intelligence (BI) intervenes, through the Systems of Support of "human decision-making". Today there are many BI solutions, being the Open Source Pentaho Business Intelligence platform one of the most used currently. This management platform covers data analysis and reporting operations, making this a flexible solution to cover our study case: "Anomaly Detection in Records of Fueling in automobiles used for illegal fuel storage in Bolivia". So, in this paper, this technological platform will be applied making some adjustments in the anomaly scores context. The principal contribution is design and development a BI solution responsible of analyze in large amount of records of anomalies in Bolivia and in this way to allow to make better decisions of control of fuel smuggling, having the right information in the right place at the right time.