試験
通常
試験時間: 00:00問題時間: 00:00
問題一覧
問題 54 / 65
In situations where data on rare diseases is extremely scarce, which approach is most effective for model performance evaluation and training?
①Appropriately oversample the minority classes and use weighting together with suitable evaluation metrics.
②Balance the classes by removing a large amount of data from the majority class.
③Optimize only overall accuracy, ignoring rare classes.
④Randomly shuffle the training data and perform standard cross-validation.
⑤The problem will be solved if we create large amounts of synthetic data using generative models.