Generalisaatioero

Generalisaatioero, also known as generalization error, is a concept in machine learning and statistics that refers to the difference between the performance of a model on training data and its performance on unseen data. It is a critical measure of a model's ability to generalize from the data it has seen to new, unseen data.

Generalization error arises because models often fit the training data too closely, capturing not just the underlying patterns but also the noise and outliers. This overfitting leads to a model that performs well on the training data but poorly on new data. Conversely, underfitting occurs when a model is too simple to capture the underlying patterns, resulting in poor performance on both training and new data.

The goal in machine learning is to find a model that minimizes generalization error. This involves balancing the complexity of the model to avoid both overfitting and underfitting. Techniques such as cross-validation, regularization, and ensemble methods are commonly used to estimate and reduce generalization error.

Generalization error is quantified using various metrics, depending on the type of problem (e.g., mean squared error for regression, accuracy for classification). It is essential for evaluating the practical utility of a model and ensuring that it performs well in real-world applications.