الفهرس Only 14 pages are availabe for public view
 |  | from | 109 |  |  |
| | | from | 109 | | |
Abstract
The missing values problem is an old one for analysis task. The waste of data, which can result from case wise deletion of missing values, obliges to propose alternatives approaches. In the current thesis, we will use a number of alternative ways of dealing with missing data, and this thesis is an attempt to outline those approaches. A regression analysis in which the independent variable (X) has a missing observation will be illustrated and discussed. Multiple (Single) imputation as an analytic strategy for missing data will be reviewed. In addition, determining the appropriate analytic approach in the presence of incomplete observations that is a major question for data analysts will be introduced. The general plan of the thesis: This thesis considers the problem of the presence of missing observation among the independent variable. It contains six chapters as : Chapter (1) introduced a general introduction, basic definitions and notations of missing data, types of missingness mechanisms, general pattern of missing data. Chapter (2) review of literatures. Chapter (3) presented the methods of handling missing data. Chapter (4) time series definition of handling missing data problem, Measure of accuracy definitions, winsorized mean definition, and Neyman allocation method definitions will be introduced. In the second part estimation of missing data by using asymmetrical winsorized mean in a time series (methodology to determine the boundaries of exponential distribution) and a generalized this method for gamma distributions are derived. Chapter (5) the estimation of linear regression models with missing observations on both explanatory and study variable divided the set of observation in four (three) parts will be discussed. Chapter (6) statistical analysis using rainfall data of Alex city and estimate the missing data by using asymmetric winsorized mean and the mean of the series