AWS Certified Machine Learning Specialty (MLS-C01) 2025 – 400 Free Practice Questions to Pass the Exam

When class imbalance exists in a dataset, it refers to the situation where some classes are underrepresented compared to others. In such cases, if the imbalance is not addressed during model training, the model tends to learn more about the majority class because it has more examples to learn from. As a result, when making predictions, the model is likely to skew its predictions toward the majority class. This means that it may classify most instances as belonging to the dominant class, leading to poor performance on the minority class.

For example, if a dataset has 90% of instances from class A and only 10% from class B, a model that predicts all instances as class A could still achieve a high accuracy (90%) but would fail entirely at recognizing class B. Thus, the consequence of not addressing the class imbalance manifests as skewed predictions favoring the majority class, which undermines the overall usefulness of the model, especially in applications where minority classes hold significant importance.

Understanding and addressing class imbalance is crucial for developing robust machine learning models that make fair and accurate predictions across all classes present in the data.