الفهرس 
InteroductionOnly 14 pages are availabe for public view
 |  | from | 118 |  |  |
| | | from | 118 | | |
Abstract
Discovering outlier values has recently become one of the most important topics of statistical inference. Outlier values (Barnett & Lewis (1994)) have been defined as the existence of an observation – or a set of observations - incompatible with the rest of the data set. These outlier values result from a human error, a system error, or a natural deviation in society. There are many ways of detecting and tackling the existence of anomalous values in the data.
The research aims
Use the Bayes method to predict the confidence limits of the exponential distribution data in the case of extreme values of θ + δ and θ*δ where θ is parameter and δ is outlier.
Use The Bootstrap method to detect anomalous values as one of the methods of detecting outliers and treating them by deletion.
Bootstrap is one of the most important methods developed in statistical inference. It deals with random samples so that we can generate several random samples from the random sample that we have. The generated random samples are pulled back from the same sample with the same or a smaller size and both have a wide range of applications. Bootstrap has been applied to previous data used in previous studies as well as to inflation data in the Egyptian economy (measured by the annual consumer price index) in the period from January 2011 to January 2018.
Developed The Bayesian method by controlling the weights given to the samples-Bayesian Bootstrap- to hide the effect of anomalous values on the sample without deleting these values. This method was applied to Gaussian distribution, non-linear distribution. It has been proven to be a prelude to knowledge of the posterior distributions of the given parameters and weights.
The research is divided into seven chapters.
The first chapter presents an introduction, definition of Outlier values, types of Outlier values, a chronological overview of the previous studies, and the purpose of the study.
The second chapter explains some of the Bayesian concepts.
The third chapter tackles the K-Nearest Neighbor (KNN) method ,gives a simple example of how it is calculated, and discusses its types for later use in the proposed method of dealing with the extreme values using the Bayes method.
The forth chapter uses the Bayes method to predict the limits of confidence for the exponential distribution using Bayes in the case of extreme values.
The fifth chapter presented the method of detecting the existence of extreme values using the Bootstrap method and the application of this method to values used in previous studies in addition to data from the Egyptian economy (such as inflation data).
The sixth chapter presents the proposed method of treating (concealing) the extreme values without deleting them using the Bayesian Bootstrap method to purify the data from the extreme values in order to know the posterior distributions of the parameters and weights used.
The study concludes
Validity of a method Bayesian Bootstrap in treating outliers values by appling to generated data that follows a normal or Gaussian distribution.
This method is also useful for applying to a non-linear distribution when adding different proportions of outliers values.
The possibility of using this method to purification data.