The “normalized” term means that each of these groupings is represented as having 1.00 samples. The “balanced” term means that each class, cluster, or topic has the same number of samples. My preferred use of confusion matrices is in balanced and normalized form. Steven Simske, in Meta-Analytics, 2019 4.7 Confusion matrix patternsĬonfusion matrices are extremely powerful shorthand mechanisms for what I call “analytic triage.” As described in Chapter 2, confusion matrices illustrate how samples belonging to a single topic, cluster, or class (rows in the matrix) are assigned to the plurality of possible topics, clusters, or classes.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |