Modeling and Classification Multicollinear Variables using Multinomial Ridge Logistic Regression Aprroach

Giatma Dwijuna Ahadi, Ismaini Zain, Santi Puteri Rahayu

Abstract

Multinomial Logistic Regression is a method used to find relationships between nominal or multinomial response variables (Y) with one or more predictor variables. Logistics Regression is a classic method that is often used to solve classification problems. Assumptions on Logistics Regression are models containing multicollinearity. Ridge Logistic Estimator (RLE) is methods to solve multicollinearity cases in Logistic Regression. Wu & Asar proposed a new ridge value that can also reduce bias in parameter estimation. Therefore, this research will discuss about Multinomial Ridge Logistic and selection the best of ridge constant values. The performance test of the ridge value will be applied to the Iris Dataset in R software. The best criteria for improvement ridge constant value by looking at the smallest standard error. The calculation results show that the Wu-Asar approach is the best ridge constant and Wald individual test shows significant results. Based on the result, show that the Wu-Asar Ridge constant value on Multinomial Ridge Logistic Regression are very good performance in estimated smaller standar error. The classification for dataset shows high results with 98% global accuracy.

Keywords: multinomial; ridge logistic regression; Wu-Asar; standard error; classification

References

J. Han, M. Kamber and J. Pei, Data Mining Concepts and Technique, 3 ed. Waltham: Morgan Kaufmann, 2012.

A. Agresti, An Intoduction to Categorical Data Analysis. New Jersey: John Wiley & Sons, 2002.

X. Yan and X. G. Su, Linear Regression Analysis : Theory and Computing. Singapore: World Scientific, 2009.

G. Khalaf and M. Iguarnane, "Ridge regression and ill-conditioning," Journal of Modern Applied Statistical Methods, vol. 13, no. 2, pp. 355-363, 2014, doi:10.22237/jmasm/1414815420

R. Schaefer, L. Roi and R. Wolfe, "A ridge logistic estimator," Communications in Statistics - Theory and Methods, vol. 13, no.1, pp. 99-113, 1984, doi: https://doi.org/10.1080/03610928408828664

J. Wu and Y. Asar, "On almost unbiased ridge logistic estimator for the logistic regression," Hacettepe Journal of Mathematics and Statistics, vol. 45, no. 3, 2016, doi: 10.15672/HJMS.20156911030

C. Nisa and S. H. Hastuti, "Kajian simulasi perbandingan metode ridge regression dan adjusted ridge regression untuk penanganan multikolinearitas," Jurnal Gaussian, vol. 12, no. 3, pp. 330-339, 2023, doi: https://doi.org/10.14710/j.gauss.12.3.330-339.

L. A. A. Sari, Pendugaan Parameter Model Regresi Logistik dengan Maximum Likelihood dan Ridge Logistic Estimator, Bogor: Institut Pertanian Bogor, 2018.

D. M. Putra and V. Ratnasari, "Pemodelan indeks pembangunan manusia (IPM) Provinsi Jawa Timur dengan menggunakan metode regresi logistik ridge," Jurnal Sains dan Seni ITS, vol.4, no.2, pp. 175-180, 2015.

E. Roflin, F. Riana, E. Munarsih, Pariyana and I. A. Liberty, Regresi Logistik Biner dan Multinomial, NEM, 2023.

N. Draper and H. Smith, Applied Regression Analysis : Third Edition, 3rd ed. Canada: John Wiley & Sons, 1998.

R. A. Johnson and D. W. Wichern, Applied Multivariate Statistical Analysis, 6 ed. New Jersey: Pearson Prantice Hall, 2007.

F. Rahmawati and R. Y. Suratman, "Performa regresi ridge dan regresi lasso pada data dengan multikolinieritas," Leibniz Jurnal Matematika, vol. 2, no. 2, pp. 1-10, 2022.

R. Amalah, A. K. Jaya and N. Sirajang, "Pemodelan geographically weighted logistic regression dengan metode ridge," ESTIMASI: Journal of Statistics and Its Application, vol. 4, no. 2, pp. 130-143, 2023. doi: https://doi.org/10.20956/ejsa.v4i2.12250.

D. W. Hosmer and S. Lemeshow, Applied Logistic Regression: Second Edition. New York: John Wiley & Sons, 2000.

F. M. Sari, K. A. Notodiputro and B. Sartono, "Analisis tingkat kemiskinan di Provinsi Sumatera Barat melalui pendekatan regresi terkendala (ridge regression, lasso, dan elastic net)," Statistika, vol. 21, no. 1, pp. 29-36, 2021, doi: https://doi.org/10.29313/jstat.v21i1.7836

DOI: https://doi.org/10.13057/ijas.v8i1.85795

Refbacks

  • There are currently no refbacks.