Aplicação da metodologia CRISP-DM: um estudo sobre diabetes utilizando bases de dados públicas

Abstract

The main objective of this study is to explore the effectiveness and relevance of the CRISP-DM model in the analysis of diabetes-related data, with the aim of developing predictive models that contribute to the understanding of the disease and its complications. Initially, a literature review on the use of CRISP-DM in the literature was carried out, highlighting its flexibility and ability to adapt to different projects and datasets. This phase allowed a detailed understanding of the structure and stages of the model, aiming at its practical application in the context of diabetes. Next, public datasets on diabetes were analyzed, coming from various sources, such as clinical care, quality registries and administrative data. Each source was evaluated for its main advantages and disadvantages, taking into account aspects such as data quality, representativeness and accessibility. This detailed analysis provided a comprehensive view of the available information and guided the selection of the most appropriate datasets for the study. During the analysis process, a documentation of the phases of the CRISP-DM model was prepared, adapted to the specific needs and requirements of the case study. This documentation not only described each step of the process, from understanding the business to implementing the models, but also provided practical guidance for its execution. In addition, it served as an educational tool, facilitating the understanding of the technical details of the study and promoting the dissemination of knowledge about data mining. The results obtained from the use of CRISP-DM were analyzed and interpreted in relation to the objectives established at the beginning of the study. The predictive models developed represent tools that can assist in the identification of disease risks, in the implementation of diabetes prevention and control strategies, and in the application of a process model to guide the process of building classifiers. In conclusion, this study demonstrated that the CRISP-DM model is an effective and comprehensive approach for the analysis of diabetes-related data. Its application allowed a deeper understanding of the patterns and risk factors associated with the disease, contributing significantly to the advancement of knowledge in the health area and data mining processes. The success of this project validates the importance of CRISP-DM as a powerful tool to drive future research and innovation in the field of data mining and public health.