Classification Performance Evaluation from Multilevel Logistic and Support Vector Machine Algorithms through Simulated Data in Python

This paper analyzes the performance of multilevel logistic and support vector machine algorithms when the objective is the stratification of the sample into two groups for binary classification. Under the data simulation in Python, we show that multilevel logistic models cannot correctly classify ob...

Full description

Saved in:
Bibliographic Details
Published inProcedia computer science Vol. 214; pp. 511 - 519
Main Authors Fávero, Luiz Paulo, Belfiore, Patrícia, Santos, Helder Prado, dos Santos, Marcos, de Araújo Costa, Igor Pinheiro, Junior, Wilson Tarantin
Format Journal Article
LanguageEnglish
Published Elsevier B.V 2022
Subjects
Logistic Models
Multilevel Models
Python
Simulation
Support Vector Machine
Logistic Models
Multilevel Models
Support Vector Machine
Simulation
Python
Online AccessGet full text
ISSN1877-0509
1877-0509
DOI10.1016/j.procs.2022.11.206

Cover

More Information
Summary:This paper analyzes the performance of multilevel logistic and support vector machine algorithms when the objective is the stratification of the sample into two groups for binary classification. Under the data simulation in Python, we show that multilevel logistic models cannot correctly classify observations under certain non-linear conditions, even when defined contextual hierarchical groups and support vector classifiers generate better predictions. Python codes are provided for replication purposes.
ISSN:1877-0509
1877-0509
DOI:10.1016/j.procs.2022.11.206