Evaluation of Naive and Covariance Algorithms in Generalized Linear Models (GLMs)
Khalilah Nurfadilah (a*), Khairil Anwar Notodiputro (b), Bagus Sartono (b)

a) UIN Alauddin Makassar
JL. HM Yasin Limpo No. 36 Makasar, Indonesia
* khalilahnurfadilah1202[at]gmail.com
b) Departemen Statistika, Fakultas Matematika dan Ilmu Pengetahuan Alam, Institut Pertanian Bogor
Institut Pertanian Bogor, Jalan Raya Dramaga, Bogor, 16680, Indonesia

Abstract

One of the important aspect in modeling is the simplicity of the model itself. Simplification of the model can be done by several methods, including the Ridge regression, LASSO and Elastic Net. In selecting the variables in these models an algorithm is developed, namely Naive and Covariance. Previous research revealed that the Covariance algorithm is superior in terms of time compared to the Naive algorithm. This is then evaluated by applying models and algorithms to male sex behavior data with the criteria of goodness, namely the simplicity of the model and the minimum AIC value. Based on the results of the study, it was found that the Covariance algorithm still outperformed the Naive algorithm in all three models.

Keywords: Covariance- Elastic Net- LASSO- Naive, Ridge Regression1

Topic: Mathematics