Michal Cerny
Computational Aspects of Regression Analysis of Interval Data
1469 - 1476
2011
5
9
International Journal of Mathematical and Computational Sciences
https://publications.waset.org/pdf/6479
https://publications.waset.org/vol/57
World Academy of Science, Engineering and Technology
We consider linear regression models where both input data (the values of independent variables) and output data (the observations of the dependent variable) are intervalcensored. We introduce a possibilistic generalization of the least squares estimator, so called OLSset for the interval model. This set captures the impact of the loss of information on the OLS estimator caused by interval censoring and provides a tool for quantification of this effect. We study complexitytheoretic properties of the OLSset. We also deal with restricted versions of the general interval linear regression model, in particular the crisp input &ndash; interval output model. We give an argument that natural descriptions of the OLSset in the crisp input &ndash; interval output cannot be computed in polynomial time. Then we derive easily computable approximations for the OLSset which can be used instead of the exact description. We illustrate the approach by an example.
Open Science Index 57, 2011