Classification

Classification is a machine learning task in which a model assigns an input to one or more predefined categories. An email filter, for example, may classify a message as spam or legitimate, while medical software may identify which condition best matches a scan. During supervised training, the model studies labeled examples and learns patterns that distinguish each class. Its performance is then measured on unseen data using metrics such as accuracy, precision, recall, and F1 score. Classification can be binary, multiclass, or multilabel depending on how many outcomes are possible and whether several labels may apply at once. Reliable systems also account for imbalanced data, uncertain predictions, changing conditions, and the cost of different mistakes.