Predição de Diabetes Utilizando Modelos de Machine Learning

This study investigates the application of supervised learning algorithmsto predict the incidence of diabetes, leveraging easilyaccessible data such as demographic information and health indicatorsto identify the most effective approaches for early diseasedetection. The primary goal is to compare the performance of differentclassification models in this task.The experimental results considered three classification methods,namely: K-Nearest Neighbors (KNN), Logistic Regression, andSupport Vector Machine (SVM). Due to the imbalanced distributionof the classes (presence or absence of diabetes), besides obtainingdecent accuracy values, the recall of all methods was highly impacted.The continuation of this work will include: i) adding moreclassification methods to the experiment, such as neural networksand ensemble-based methods; ii) compare the obtained results withthe literature; and iii) consider the impact of data pre-processingsteps to mitigate the class imbalance.