On Some Goodness-of-Fit Tests and Their Connection to Graphical Methods with Uncensored and Censored Data
In this work, we present goodness-of-fit tests related to the Kolmogorov-Smirnov and Michael statistics and connect them to graphical methods with uncensored and censored data. The Anderson-Darling test is often empirically more powerful than the Kolmogorov-Smirnov test. However, the former one cannot be related to graphical tools by means of probability plots, as the Kolmogorov-Smirnov test does. The Michael test is, in some cases, more powerful than the Anderson-Darling and Kolmogorov-Smirnov tests and can also be related to probability plots. We consider the Kolmogorov-Smirnov and Michael tests for detecting whether any distribution is suitable or not to model censored or uncensored data. We conduct numerical studies to show the performance of these tests and the corresponding graphical tools. Some comments related to big data and lifetime analysis, under the context of this study, are provided in the conclusions of this work.
KeywordsAnderson-Darling Kolmogorov-Smirnov and Michael tests Big data Censored data Test power
The authors thank the editors and reviewers for their comments on this manuscript. This research work was partially supported by FONDECYT 1160868 grant from the Chilean government.
- 3.Aykroyd, R.G., Leiva, V., Ruggeri, F.: Recent developments of control charts, identification of big data sources and future trends of current research. Technological Forecasting and Social Change (pages in press) (2019)Google Scholar
- 22.Meatless, M., Rousseeuw, P.J., Croux, C., et al.: robustbase: Basic Robust Statistics. R package version 0.93-3 (2018). http://robustbase.r-forge.r-project.org