Understanding classification model jargon

As with regression, classification problems come with their own set of jargon. There is some overlap with terms used in regression, but there are also some new terms specific to classification:

• Categories, labels, or classes: These terms are used interchangeably to represent the various distinct choices for our prediction. For example, we could have a fraud class and a not fraud class, or we could have sitting, standing, running, and walking categories.
• Binary classification: This type of classification is one with only two categories or classes, such as yes/no or fraud/not fraud.
• Multi-class classification: This type of classification is one with more than two classes, such as a classification trying to assign one of hot dog, airplane, cat, and so on, to an image.
• Labeled data or annotated data: Real-world observations or records that have been paired with their corresponding class. For example, if we are predicting fraud via transaction time, this...