Classification metrics can't handle a mix of continuous and binary targets

Classification metrics are used to evaluate the performance of models that predict categorical targets, such as binary (0/1) or multi-class (e.g. 0/1/2/3) outcomes. These metrics include accuracy, precision, recall, F1-score, and others.

However, if the target variable contains a mix of continuous and binary values, classification metrics cannot be used to evaluate the model's performance. In such cases, regression metrics such as mean squared error (MSE), mean absolute error (MAE), and R-squared should be used instead.

For example, if the target variable represents the probability of an event occurring, a continuous value between 0 and 1, regression metrics can be used to evaluate the model's performance in predicting this probability. If the target variable represents a binary outcome, classification metrics are appropriate.