Degree
Master of Science in Data Science
Department
Department of Computer Science
Faculty/ School
School of Mathematics and Computer Science (SMCS)
Date of Submission
Spring 2023
Supervisor
Dr. Tahir Syed, Assistant Professor, Department of Computer Science, School of Mathematics and Computer Science (SMCS)
Co-Supervisor
Behraj Khan, Lecturer, Department of Computer Science, School of Mathematics and Computer Science (SMCS), Institute of Business Administration (IBA), Karachi
Keywords
Prior Probability Shift, Label Shift Detection, Imbalance Learning, Prior Probability Shift Adaptation, Label Shift Datasets
Abstract
The extra-sample performance of a learner may deteriorate over time due in part to factors outside of the learning process and therefore may not be corrected by better training regimes. One such example is prior probability shift, where P(Y) differs at test time. This discrepancy is often observed in the study of the phenomenon of long tails in P(Y) at training time, leading to training enhancement with no change to the way the learnt model is assessed. This occurrence is frequently associated with benchmarks to class imbalance. In this survey, we will investigate the extent of the problem, explore the learning settings, and propose and evaluate changes to the latter. This study categorizes existing strategies for handling prior probability shift, provides an overview of representative techniques and algorithms, discusses evaluation methodology, and presents illustrative applications. The aim is to offer a comprehensive introduction to prior probability shift adaptation for researchers and industry practitioners.
Document Type
Restricted Access
Submission Type
Research Project
Recommended Citation
Saleem, S. (2023). A Survey on Prior Probability Shift Adaptation (Unpublished graduate research project). Institute of Business Administration, Pakistan. Retrieved from https://ir.iba.edu.pk/research-projects-msds/22
The full text of this document is only accessible to authorized users.
