Techniques for Designing Classification Problems
Author | : Ratna Raju Mukiri |
Publisher | : Educreation Publishing |
Total Pages | : 197 |
Release | : 2018-10-30 |
Genre | : Education |
ISBN | : |
Download Techniques for Designing Classification Problems Book in PDF, Epub and Kindle
The thesis is mainly concerned with classification of tasks and related issues that appear in real-world scenarios, such as incomplete records and irrelevant and/or redundant pieces of information, imbalanced class distribution and imbalanced error costs. There is no universally accepted best classifier and general rules for appropriate metric to select in a certain context exist. Translating the data characteristics and problem goals into appropriate performance, metrics, appropriate classifier is successful data mining process. An original meta-learning framework for automated classifier selection is presented in the case studies section for baseline performance assessment.