TUNING PARAMATER SELECTION IN PENALIZED LOGISTIC REGRESSION WITH APPLICATION IN CANCER

Year 2019, Volume: 18 Issue: 36, 11 - 22, 31.12.2019

Sahar Fadhil Mohammed Al-khateeb

Abstract

Variable selection is an important subject in regression analysis intended to select the best subset of
predictors. In cancer classification, gene selection plays an important issue. The Least Absolute Shrinkage and
Selection Operator (LASSO) is one of most used penalized method. In logistic regression, Lasso right the
traditional parameter estimation method, maximum log-likelihood, by adding the L1-norm of the parameters to
the negative log-likelihood function. Lasso depends on the tuning parameter. Finding the optimal value for the
tuning parameter is one of the most important topics. There are three popular methods to select the optimal
value of the tuning parameter: Bayesian Information Criterion (BIC), Akaike Information Criterion (AIC), and
Cross-Validation (CV). The aim of this paper is to evaluate and compare these three methods for selecting the
optimal value of tuning parameter in terms of coefficients estimation accuracy and variable selection through
simulation studies and application in cancer classification. 

Keywords

Kanser sınıflandırması, Gen seçimi. Lasso, cezalandırılmış Lojistik regresyon

References

• Abdalteef A. M., (2018), “Variable selection in Poisson regression model usingpenalizedlikelihood methods”, University of Mosul ,Faculty of Mathematics andStatisti,Master Thesis in Statistics, Mosul.Adragni, K. P., (2014), Independent screening in high-dimensional exponential familypredictors space, Journal of Applied Statistics, 42(2), 347–359.Algamal Z, Hisyam M, (2015), “Penalized logistic regression with the adaptive LASSOfor gene selection in high-dimensional cancer classification”,Expert Systems withApplications;42(23):9326-9332.Algamal, Z. Y. and Lee, M. H., (2015),“Regularized logistic regression with adjustedadaptive elastic net for gene selection in high dimensional cancer classification”,Computers in Biology and Medicine, 67, 136-145.Algamal, Z. Y., (2016),“ Adaptive Penalized Likelihood Methods In High Dimensionalgeneralized Linear Models", Unpublished, phD Thesis, UniversitiTeknologi Malaysia.Androulakis, E., Koukouvinos, C. and Mylona, K., (2011), Tuning parameter estimationin penalized least squares methodology. Communications in Statistics - Simulation andComputation. 40(9), 1444–1457.Azhaar, J. K ., (2014), “Multivariate Data Analysis for Diagnosis of phthalmic DiseasesUsing the Distributive Function and Logistic Regression Comparative Study",Mustansiriya University, Faculty of Management and Economics, Master Thesis inStatistics, BaghdadFan, Y. and Tang, C. Y., (2013), Tuning parameter selection in high dimensionalpenalized likelihood, Journal of the Royal Statistical Society. Series B (Methodological).75(3), 531–552.Friedman, J., Hastie, T. and Tibshirani, R., (2010), “Regularization paths for generalizedlinear models via coordinate descent. Journal of Statistical Software. 33(1), 1–22.Kar, S., Das Sharma, K. and Maitra, M. (2015). Gene selection from microarray geneexpression data for classification of cancer subgroups employing PSO and adaptive Knearest neighborhood technique. Expert Systems with Applications. 42(1), 612– 627.Tibshirani, R. (1996). Regression shrinkage and selection via the lasso. Journal of theRoyal Statistical Society. Series B (Methodological). 58(1), 267–288.Park, B. U. and Hastie, T. (2007). L1-regularization path algorithm for generalized linearmodels. Journal of the Royal Statistical Society. Series B (Methodological). 69, 659677.Mkhadri, A. and Ouhourane, M. (2015). A group VISA algorithm for variable selection.Statistical Methods & Applications. 24, 41–60.