The Estimation of
The Estimation of
Regression Models with Censored Data Using Logistic and Tobit Models
Atefeh younesi a, Elham Kamangar b
a Department of Mathematics, Master of Science,
Zanjan University, Zanjan, Iran
b Department of Mathematics, Payam Noor University of Tehran
Abstract: In
statistics, censoring occurs when the value of an observation is only partially
known. The aim of this paper is estimation of a regression model with censored
data using Logistic and Tobit models. We have compared two models based on
goodness of fit and forecasting accuracy criteria. We have used the data of
rate of return and volatility of Tehran Stock Exchange. Results indicate that
Based on Akaike info criterion, Schwarz criterion, Hannan-Quinn criterion and
Log likelihood, the model of Tobit has better goodness of fit than Logistic
model. Criteria of RMSE and MAE indicate that the Tobit model has more accuracy
of forecasting than Logistic Model.
[Atefeh
younesi, Elham Kamangar. The
Estimation of Regression Models with Censored Data Using Logistic and Tobit
Models. Life Sci J 2013;10(7s):545-550]
(ISSN:1097-8135). http://www.lifesciencesite.com. 85
Keywords: Regression
Models, Censored Data, Logistic, Tobit.
