Uma comparação das abordagens para tratamento de concept drifts em dados processados com machine learning

Abstract

This work presents a comparison of approaches for handling concept drift in machine learning models applied to processed data. Concept drift refers to changes in the data distribution over time, which can impact the performance of machine learning models. The objective of this study is to analyze strategies for dealing with concept drift and evaluate their effectiveness in maintaining model performance over time. The considered approaches include retraining, ensemble-based retraining, and adaptive algorithms. The methodology involved implementing and experimenting with these approaches on a real dataset, as well as conducting comparative analyses of the results. The findings indicate that each approach has its own advantages and disadvantages depending on the scenario and data characteristics, and there is no one-size-fits-all solution. This comparison provides valuable insights for the appropriate selection of concept drift handling techniques in machine learning problems.