KERNEL SMOOTHING AS AN IMPUTATION TECHNIQUE FOR RIGHT-CENSORED DATA

Year 2020, , 1 - 6, 27.11.2020

Dursun Aydın , Ersin Yılmaz

https://doi.org/10.18038/estubtda.817979

Abstract

Imputation of right-censored observations is an important problem in statistics and other applied sciences. Since right-censored data sets are common in medical studies and survival analysis, researchers should be careful about data quality. In this sense, imputation techniques are used to correctly estimate and complete censored data points. This study introduces the kernel smoothing method as an imputation method that takes into account the structure of the data and the individual effects of the accessible data points with kernel weights. The basic idea is to obtain a nonparametric model from the missing data set and consider sample predictions to estimate the censored ones. A simulation study is conducted to show the benefits of the method, and it is also compared with Ordinary Least Squares (OLS) based imputation, which is one of the widely used imputation methods and works similar to the proposed method.

Keywords

Kernel smoothing, right-censored data, imputation technique, OLS imputation

References

• Ahmed, S.E., Aydın, D. and Yılmaz, E. (2020). Nonparametric regression estimates based on imputation techniques for right-censored data, ICMSEM 2019: Proceedings of the Thirteenth International Conference on Management Science and Engineering Management, 109-120.
• Aydın, D., Yılmaz, E. (2017). Modified Spline Regression Based On Randomly Right-Censored Data: A Comparison Study, Communications in Statistics-Simulation and Computation, doi: 10.1080/03610918.2017.1353615.
• Aydın D. and Yılmaz E. (2016). Right-censored nonparametric regression: A comparative simulation study, TEM Journal, 5(4), 446-450. Buckley, J., James, I. (1979). Linear Regression with Censored Data. Biometrika, Vol. 66(3), 429-436.
• Chen, L. and Sun, J. (2010). A multiple imputation approach to the analysis of interval-censored failure time data with the additive hazards model, Comput. Stat. Data. Anal., 54(4), 1109-1116.
• Hasler, C. and Craiu, R.V. (2020). Nonparametric imputation method for nonresponse in surveys, Statistical Methods & Applications, 29, 25-48.
• Koul, H., Susarla, V., Van Ryzin, J. (1981). Regression Analysis with Randomly Right-Censored Data. The Annals of Statistics, 1276-1285.
• Leurgans, S. (1987). Linear models, random censoring and synthetic data. Biometrika Vol. 74, 301-309.
• Musil, C.M., Warner, C.B., Yobas, P. K. and Jones, S.L. (2002). A comparison of imputation techniques for handling missing data, Western Journal of Nursing Research, 24(7), 815-829.
• Miller, R. G. (1976). Least squares regression with censored data. Biometrika Vol.63, 449-64.
• Nadaraya, E. A., (1964). On Estimating Regression. Theory of Probability and Its Applications, Vol.10, 186-190.
• Stute, W. (1993). Consistent Estimation under Random Censorship When Covariables are Present. Journal of Multivariate Analysis, Vol.45, 89-103.
• Watson, G. S. (1964). Smooth regression analysis. Sankhya A 26, 359-72.
• Wei, G.C.G. and Tanner, M.A. (1991). Applications of multiple imputation to the analysis of censored regression data, Biometrics, 47(4), 1297-1309.