الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 155 |  |  |
| | | from | 155 | | |
Abstract
In this thesis, we introduce new generalized distributions by using two approaches. The first one is called the Marshall and Olkin method, which enables us to add a parameter to any family of distributions. We applied this method on Makeham and Burr distributions. We investigated some reliability properties of a flexible extended family of distributions. Quantiles are obtained and a probability distribution has for particular parameter values a bathtub hazard rate function. Stochastic ordering and limiting distributions of extreme order statistics are verified. The second approach depends on the definition of the interested random variable itself. We focused our attention on the slash random variable, which is defined as the ratio of two independent normal and uniform (on the interval (0,1)) random variables. Since the beta distribution with the two parameters α and β on the interval (0,1) reduces to the uniform distribution when α=β=1, then our main idea was to replace the uniform random variable in the denominator of the slash and skew slash random variable by the beta random variable with the parameters α and β. Hence we introduced new families of univariate multivariate generalized slash distribution and skew slash distribution as the scale mixture of the normal, exponential power, skew normal and the beta distributions. It is shown that the new generalized families of distributions have heavier tails than the known slash and skew slash distributions. Furthermore, Moments and the invariant property under linear transformations are addressed. The simulation and the fitting of a real data ensure that our proposed generalized models are flexible and better than the pervious models in the literature.