| Title |
Multiple outliers identification method in accelerated failure time (aft) - regression models / |
| Translation of Title |
Daugybinių išskirčių identifikavimo metodas pagreitinto gedimų laiko (AFT) - regresijos modeliams. |
| Authors |
Sinkevičius, Deivydas |
| Full Text |
|
| Pages |
48 |
| Keywords [eng] |
outliers, accelerated failure time (AFT) regression, BP method, Davies Gather method, robust estimation |
| Abstract [eng] |
The purpose of the study is to give modifications of outliers identification methods for normal linear regression to more general case of accelerated failure time (AFT) regression and compare them. In this work three outlier search methods are investigated: BP and David Gather (DG) based on one of two parameter estimation methods: ordinary least squares and robust. The objective of the first part is to modify BP outlier search method for accelerated failure time (AFT) regression models. The objective of the second part is to compare all three methods using data generation and to provide practical examples. In many situations, BP outlier search method identifies outliers better then both DG methods (ordinary least squares and robust). Analysis of real data examples confirms simulation results. Obtained results proved, that BP outlier search method can be useful for outliers search in accelerated failure time (AFT) regression model. |
| Dissertation Institution |
Vilniaus universitetas. |
| Type |
Master thesis |
| Language |
English |
| Publication date |
2021 |
