The Impact of Decision Cut Points on the Recovery of Low-Probability Events

• In Event: Hilton Roundtable Session 3
  In Roundtable Session: Collection of Modern Statistics

Abstract

This study will attempt to determine if the manipulation of decision cut points impacts the recovery of low probability events. Decision cut points are used in the classification process to determine class ownership, with the default decision cut point being 0.50. Manipulating decision cut points involves moving the above-mentioned default value in order to maximize a pre-determined metric. Results follow previous research that states the impact of decision cut points are largely case specific. Results also indicate that decision cut points are most successful in the recovery of low probability events when cut point values are lower.

Authors

• Kendall Smith, Ball State University

• Jocelyn H. Bolin, Ball State University