Class imbalance is a human-made concept describing when the class distribution in a classification dataset is skewed — one or more minority classes occur significantly less frequently than majority classes. It is defined by three parameters: (1) the number of classes, (2) the count or proportion of samples per class, and (3) the imbalance ratio (majority count divided by minority count), above which the distribution is deemed imbalanced. It persists through formal definitions in data science textbooks, evaluation metric suites (precision, recall, F1) designed to remain informative under skew, and algorithmic techniques (resampling, cost-sensitive learning) that address the operational consequences of skewed distributions on model training. [formal: classis inaequalis | substrate: behavior | horizon: a life | explicit: yes | epoch: 0.01]
Accepted ontology entry
class imbalance
Class imbalance is a human-made concept describing when the class distribution in a classification dataset is skewed — one or more minority classes occur significantly less frequently than majority classes. It is defined by three parameter…
Definition
Why it is in scope
A human-made concept in machine learning and statistics describing a situation where one or more classes in a classification dataset are significantly underrepresented relative to others, creating skewed class distributions that affect model training and evaluation.
Names and aliases
- class imbalanceen · CANONICAL
Relations from this entry
- cmrmi0v8g00bwd1nlrx23qz6gDEPENDS_ON →
DEPENDS_ON removal test: class imbalance is defined as a skewed class distribution within a classification dataset. Remove the concept of classification and the concept of class imbalance ceases to operate — it has no meaning outside of classification. It is not merely sayable without classification; it is inconceivable.
- cmrg0scos00ef2a1nklfvbk7xDEPENDS_ON →
Class imbalance is a specific problem phenomenon that only exists within the context of machine learning. Remove ML and the concept of class imbalance — as the imbalance of training examples across classes — ceases to operate. The class distribution itself is a data property; class imbalance as a named phenomenon requiring algorithmic attention is a ML-specific construct. Removal test: without ML, there is no learning system to be imbalanced.
Relations to this entry
No accepted relations in this direction.
Record identity
- Created
- Aug 5, 2026, 7:14 AM UTC
- Content hash
- 3c3fa8070c22e0b708380891d290f60cdffc409722443b02e88ef17f5e33d665