Which type of prediction would you use to find out if a student graduates on time?

Using binary prediction is appropriate for determining whether a student graduates on time, as this type of prediction focuses on outcomes with two distinct categories: in this case, "graduated on time" and "did not graduate on time." This binary classification allows you to categorize each student into one of these two possible outcomes based on relevant features or data points, such as grades, attendance, and other academic metrics.

Binary prediction is particularly useful in scenarios where a yes/no answer is required, making it ideal for situations involving graduation status. The simplicity of this model makes it easy to interpret and actionable, as the results directly inform whether interventions may be needed to help a student graduate on time or not.

Other types of predictions mentioned are less applicable in this context. Numeric predictions typically deal with predicting quantities, continuous predictions involve outcomes that can fall anywhere on a continuous scale, and categorical predictions might include multiple classes beyond a simple binary outcome. Thus, binary prediction is the most fitting choice for assessing on-time graduation status.