Sobreamostragem Técnica refers to a collection of methods used in pré-processamento de dados to address the issue of desequilíbrio de classes in datasets, particularly for classification tasks in machine learning. Class imbalance occurs when the number of instances of one class significantly outweighs the number of instances of another class, which can lead to biased models that favor the majority class.
Oversampling techniques work by artificially increasing the representation of the classe minoritária. This can be achieved through various methods, such as:
- Sobreamostragem Aleatória: This method involves randomly duplicating examples from the minority class until the desired balance with the majority class is achieved. While simple, it can lead to overfitting pois ela replica os mesmos exemplos.
- SMOTE (Técnica de Oversampling de Minorias Sintéticas): SMOTE generates synthetic examples by interpolating between existing minority class instances. This helps create a more generalized model as it introduces variability rather than merely duplicating data.
- ADASYN (Amostragem Sintética Adaptativa): This technique is similar to SMOTE but focuses on generating more synthetic examples in regions of the espaço de características where the minority class is less dense, providing a more adaptive approach to oversampling.
Oversampling can improve the performance of classifiers by providing more balanced training data, which can lead to better generalization and accuracy for the minority class. However, it is important to evaluate the model’s performance using metrics that consider class balance, such as F1-score, precision, and recall, to ensure that the oversampling technique is effectively addressing the imbalance issue.