In the realm of machine learning, variance refers to the extent to which a model's predictions are sensitive to alterations in the training data. A high-variance model is overly complex and exhibits overfitting to the training data. This implies that the model learns not only the underlying patterns but also the noise and unique characteristics of the training data. Consequently, such a model demonstrates good performance on the training data, yet performs poorly when dealing with unseen data or the test set.
Conversely, a low-variance model is too simplistic and suffers from underfitting the training data. It fails to effectively capture the relevant features and patterns within the data, resulting in subpar performance on both the training and test sets.