In this research work an attempt has been made to investigate about the Robustness of the Bayesian Information criterion to estimate the order of the autoregressive process when the error of this model,  Submits to a specific distributions and different cases of the time series on various size of samples by using the simulation,  This criterion has been studied by depending on ten distributions, they are (Normal, log-Normal, continues uniform, Gamma , Exponential, Gamble, Cauchy, Poisson, Binomial, Discrete uniform) distributions, and then it has been reached to many collection and recommendations related to this object , when the series residual variable is subject to each  ( Poisson , Binomial , Exponential , Dis

In this paper, we made comparison among different parametric ,nonparametric and semiparametric estimators for partial linear regression model users parametric represented by ols and nonparametric methods represented by cubic smoothing spline estimator and Nadaraya-Watson estimator, we study three nonparametric regression models and samples sizes  n=40,60,100,variances used σ²=0.5,1,1.5 the results  for the first model show that N.W estimator for partial linear regression model(PLM) is the best followed the cubic smoothing spline estimator for (PLM),and the results of the second and the third model show that the best estimator is C.S.S.followed by N.W estimator for (PLM) ,the

Semiparametric methods combined parametric methods and nonparametric methods ,it is important in most of studies which take in it's nature more progress in the procedure of accurate statistical analysis which aim getting estimators efficient, the partial linear regression model is considered the most popular type of semiparametric models, which consisted of parametric component and nonparametric component in order to estimate the parametric component that have certain properties depend on the assumptions concerning the parametric component, where the absence of assumptions, parametric component will have several problems for example multicollinearity means (explanatory variables are interrelated to each other) , To treat this problem we use

In this research we study a variance component model, Which is the one of the most important models widely used in the analysis of the data, this model is one type of a multilevel models, and it is considered as linear models , there are three types of linear variance component models ,Fixed effect of linear variance component model, Random effect of linear variance component model and Mixed effect of linear variance component model . In this paper we will examine the model of mixed effect of linear variance component model with one –way random effect ,and the mixed model is a mixture of fixed effect and random effect in the same model, where it contains the parameter (μ) and treatment effect (τ_i ) which  has

In this paper, compared eight methods for generating the initial value and the impact of these methods to estimate the parameter of a autoregressive model, as was the use of three of the most popular methods to estimate the model and the most commonly used by researchers MLL method, Barg method  and the least squares method and that using the method of simulation model  first order autoregressive through the design of a number of simulation experiments and the different sizes of the samples.

This research dealt with the analysis of murder crime data in Iraq in its temporal and spatial dimensions, then it focused on building a new model with an algorithm that combines the characteristics associated with time and spatial series so that this model can predict more accurately than other models by comparing them with this model, which we called the Combined Regression model (CR), which consists of merging two models, the time series regression model with the spatial regression model, and making them one model that can analyze data in its temporal and spatial dimensions. Several models were used for comparison with the integrated model, namely Multiple Linear Regression (MLR), Decision Tree Regression (DTR), Random Forest Reg

In this paper, we will study non parametric model when the response variable have missing data (non response) in observations it under missing mechanisms MCAR, then we suggest Kernel-Based Non-Parametric Single-Imputation instead of missing value and compare it with Nearest Neighbor Imputation by using the simulation about some difference models and with difference cases as the sample size, variance and rate of missing data.