AI Fundamentals

Class imbalance arises when the training dataset contains a disproportionate number of instances of one class compared to another. Because most data mining algorithms are designed to optimize overall accuracy, they tend to become biased toward the majority class, causing the model to perform poorly and produce higher error rates when classifying the minority class. This is a well-known data quality issue that directly affects model performance, unlike dimension reduction (a technique used to simplify data and often improve accuracy) or digital tracking (unrelated to training data composition).