In this paper, previous studies about Fuzzy regression had been presented. The fuzzy regression is a generalization of the traditional regression model that formulates a fuzzy environment's relationship to independent and dependent variables. All this can be introduced by non-parametric model, as well as a semi-parametric model. Moreover, results obtained from the previous studies and their conclusions were put forward in this context. So, we suggest a novel method of estimation via new weights instead of the old weights and introduce

Paper Type: Review article.

another suggestion based on artificial neural networks.

Abstract

 The problem of missing data represents a major obstacle before researchers in the process of data analysis in different fields since , this problem is a recurrent one in all fields of study including social , medical , astronomical and clinical experiments .

The presence of such a problem within the data to be studied may influence negatively on the analysis and it may lead to misleading conclusions , together with the fact that these conclusions that result from a great bias caused by that problem in spite of the efficiency of wavelet methods but they are also affected by the missing of data , in addition to the impact of the problem of miss of accuracy estimation

In this paper we estimate the coefficients and scale parameter in linear regression model depending on the residuals are of type 1 of extreme  value distribution for the largest values . This can be regard as an improvement for the studies with the smallest values . We study two estimation methods ( OLS  & MLE ) where we resort to Newton – Raphson (NR) and Fisher Scoring methods to get MLE estimate because the difficulty of using the usual approach with MLE . The relative efficiency criterion is considered beside to the statistical inference procedures for the extreme value regression model of type 1 for largest values . Confidence interval , hypothesis testing for both scale parameter and regression coefficients

In this research, some robust non-parametric methods were used to estimate the semi-parametric regression model, and then  these methods were compared using the MSE comparison criterion, different sample sizes, levels of variance, pollution rates, and three different models were used. These methods are S-LLS S-Estimation -local smoothing, (M-LLS)M- Estimation -local smoothing, (S-NW) S-Estimation-NadaryaWatson Smoothing, and (M-NW) M-Estimation-Nadarya-Watson Smoothing.

The results in the first model proved that the (S-LLS) method was the best in the case of large sample sizes, and small sample sizes showed that the

Nonlinear time series analysis is one of the most complex problems ; especially the nonlinear autoregressive with exogenous variable (NARX) .Then ; the problem of model identification and the correct orders determination considered the most important problem in the analysis of time series . In this paper , we proposed splines  estimation method for model identification , then we used three criterions for the correct orders determination. Where ; proposed method used to estimate the additive splines for model identification , And the rank determination depends on the additive property  to avoid the problem of curse dimensionally . The proposed method is one of the nonparametric methods , and the simulation results give a

Theresearch took the spatial autoregressive model: SAR and spatial error model: SEM in an attempt to provide a practical evident that proves the importance of spatial analysis, with a particular focus on the importance of using regression models spatial andthat includes all of them spatial dependence, which we can test its presence or not by using Moran test. While ignoring this dependency may lead to the loss of important information about the phenomenon under research is reflected in the end on the strength of the statistical estimation power, as these models are the link between the usual regression models with time-series models. Spatial analysis had

This paper presents a grey model GM(1,1) of the first rank and a variable one and is the basis of the grey system theory , This research dealt  properties of grey model and a set of methods to estimate parameters of the grey model GM(1,1)  is the least square Method (LS) , weighted least square method (WLS), total least square method (TLS) and gradient descent method  (DS). These methods were compared based on two types of standards: Mean square error (MSE), mean absolute percentage error (MAPE), and after comparison using simulation the best method was applied to real data represented by the rate of consumption of the two types of oils a Heavy fuel (HFO) and diesel fuel (D.O) and has been applied several tests to

The use of Bayesian approach has the promise of features indicative of regression analysis model classification tree to take advantage of the above information by, and ensemble trees for explanatory variables are all together and at every stage on the other. In addition to obtaining the subsequent information at each node in the construction of these classification tree. Although bayesian estimates is generally accurate, but it seems that the logistic model is still a good competitor in the field of binary responses through its flexibility and mathematical representation. So is the use of three research methods data processing is carried out, namely: logistic model, and model classification regression tree, and bayesian regression tree mode

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