Fraud identification architecture using data mining and machine learning in a private transport company that operates by applications

With the digital transformation over the years and the recent expansion of the use of different applications, it is possible to notice a significant change in several businesses. The diversification of electronic payments has contributed to companies suffering more from fraud. The purpose of this article is to detail a fraud detection architecture based on the identification of patterns of behavior and was applied in the racing bases of the usage of an application transport company. The study considered the construction of an artifact capable of minimizing the problem using unsupervised and supervised algorithms and machine learning techniques. The research was carried out using the DSR - Design Science Research method and considered the stages of construction of a possible conceptual structure with the systematic review of the literature, studies of fraud practices and machine learning techniques used for the detection. The architecture was implemented and allowed to validate the model capable of identifying suspected fraud in a more accurate way.