Inliers detection in pareto distribution

Abstract

Inliers in a data set are subset of observations not necessarily all zeroes, which appears to be inconsistent with the remaining data set. They are either the resultant of instantaneous or early failures usually encountered in life testing, financial, clinical trial and many other studies. Pareto distribution has recently been used as a model for fi le sizes on the internet, insurance losses, and financial behavior of the stock market and in telecommunication systems. Many of the empirical studies also use Pareto’s law for representing long tail distributions. The present article discusses the inliers detection in Pareto distribution in the context of life testing experiments where data involves instantaneous and early failures. The methods are illustrated on simulated data sets and on a real life data

Keywords:

• exponential distribution
• outliers
• labeled slippage model
• optimal estimating equations
• mixture distribution
• MLEs
• asymptotic distribution
• tests for parameters
• early failures
• Fisher information
• Schwarz’s Information criterion