The term data mining is used for the first time in the mid-nineties by Fayaad et al [13:17], and was associated at the time with the steps that must go by the establishment to pursue technical data mining; which is known today as the data mining and knowledge discovery. Then poured research to support this approach to amend the process and the introduction of new methods and adapting old methods to solve problems and reconcile models and test their credibility in light of large data sets. Has met widespread acceptance in large organizations in the West because they found that entry to this world would help achieve their goals and improve their competitive positions significantly. The most sciences that contributed to the science of data mining are statistics, machine learning and information systems. The most modern explicit methods for data mining are neural networks, decision trees, the analytic hierarchy process, nonparametric regression and analysis of symmetry. Some traditional statistical methods (like principal components analysis, factor analysis, discriminant analysis, cluster analysis, probit and logit models, nearest neighbors method, generalized additive models and mathematical programming are developed and / or used to complement modern methods of data analysis process within the framework of data mining science. It was natural to keep pace with this new software development containing these modern methods, but the use of these programs and therefore these methods in research is still in the minimalistic because the modernity of these topics and scarcity of publications in Arabic and therefore difficult to understand. So, it has declined most researchers for those methods to reconcile the relationship between the dependent variable and independent variables with the aid a multiple linear regression model for easy to understand and use. However, the modern applications proved weak credibility of multiple linear regression model in fitting most contemporary problems that characterized nonlinearity and the presence of interactions between variables due to large data sets. The aim of this research is to promote these methods whether parametric, nonparametric or semi-parametric, and meet modern applications which those methods were used successfully.