INLIERS DETECTION USING SCHWARTZ INFORMATION CRITERION
Keywords:
Exponential Distribution, Instantaneous Failures, Mixture Distribution, Maximum Likelihood, Asymptotic Distribution, Labeled Slippage Model, Early Failures, Inliers.Abstract
In failure time distributions, inliers in a data set are subset of observations sufficiently small relative to the rest of the observations, which appears to be inconsistent with the remaining data set. They are either the resultant of instantaneous failures or early failures, experienced in many life-testing experiments. The model used in outliers, where r observations are outliers is modified by)(EM r FG ∂∂ / strictly decreasing instead of increasing function of X to represent this situation, where F is the target distribution and G generates inliers. Usually number of inliers is not known and is to be determined. We use the information criterion given by Schwartz (1978) to detect the number of inliers in the model. The method is illustrated with a simulated experiment and a real life data
Downloads
References
Abramovitz, M. and Stegun, I. A. (1965). Handbook of Mathematical Functions, General
publishing Company Ltd, Canada.
Aitchison, J. (1955). On the distribution of a positive random variable having a discrete
probability Mass at the origin, J. Amer. Stat. Assn., Vol. 50, p.901-908.
Akaike, H. (1974). A new look at the Statistical identification model, IEEE Trans. Auto.
Control, 19, p.716-723.
Balakrishanan, N. and Chen, J. (2004). Detecting a change point in a sequence of extreme
value observations, J. Prob. Statist. Science, 2(1), p.55-64.
Barnett, V. and Lewis, T. (1984). Outliers in Statistical data, John Wiley & Sons, New York.
Gather, U. and Kale, B.K (1986). Outlier generating models: A review, Tech. Report No. 117,
Dept. of statistics and Acturial Mathematics, University of Iowa, Iowa city, Iowa.
Jayade, V.P. and Prasad, M.S. (1990). Estimation of parameters of mixed failure time
distribution. Comm, Statist. – Theory and Methods, 19(12), p.4667-4677.
Kale, B.K. (2003). Modified failure time distributions to accommodate instantaneous and early
failures, Industrial Mathematics and Statistics, Ed. J. C. Misra, p.623-648, Narosa Publishers,
New Delhi.
Kale, B.K. and Muralidharan, K. (2000). Optimal estimating equations in mixture distributions
accommodating instantaneous or early failures. J.Indian . Statist. Assoc., 38, p.317-329.
Kleyle , R.M. and Dahiya, R.L. (1975). Estimation of parameters of mixed failure time
distribution from censored data, Comm. Statist. – Theory and Methods, 4(9), p.873-882.
Muralidharan, K. (1999). Tests for the mixing proportion in the mixture of a degene-rate and
exponential distribution, J. Indian Stat. Assn., Vol. 37, issue 2, p.105-119.
Muralidharan, K. (2000). The UMVUE and Bayes estimate of reliability of mixedfailure time
distribution., Comm. Statist- Simulations & Computations, 29(2), p. 603-619.
Muralidharan, K. and Kale, B.K. (2002). Modified Gamma distribution with Singularity at
zero. Comm. Statist- Simulations & Computations, 31(1), p.143-158.
Muralidharan and Lathika (2004). The Concept of inliers. Proceedings of first Sino-
International Symposium on Probability, Statistics and Quantitative Management., Taiwan, October,
p. 77-92.
Schwartz (1978). Estimating the dimensionality of a model. The Annals of Statistics, Vol.
(2), p. 461-464.
Vannman. K. (1991). Comparing samples from nonstandard mixtures of distributions with
Applications to quality comparison of wood. Research report 1991 : 2 submitted to Division of
Quality Technology, Lulea University, Lulea, Swedon.
