The Regression analysis is considered as a basic element of statistics science elements, and it is an important style of applied statistics when studying different social and economical phenomena. As well as it is considered the most statistical method used in different sciences and fields, and it determines in clear picture the relations among the variables in the form of equation replaced from the degree of parametric on the strength and the importance of this relationship .And ;also it repots the degree of prediction and response ;also ,it is necessary in regression planning and making decisions for finding regression equation .And it is one of modern styles which tak observable caring especially in artificial neural networks. Our main aim is using neural networks in regression in special visage comparison of neural networks styles with classical styles in analyzing linear and non linear regression of prediction in monthly production of electric energy for the seven years i .e from 2006 to 2012 . The research includes the method of analyzing prediction and regression by using the two methods, classic and neural networks and making comparison between the two methods , the 1st method involve the linear model by using the data of chemical operation as a dependent variable 7 reaction time ,temperature as an independent variables for 2 years while the 2nd used the data for seven last years to reach the best method in evaluating statistical parameters , to support the research we generate three types of data one for simple linear regression , 2nd for time series and 3rd for non liner data. The results prove that neural networks are the best method in non linear models while the reverse in the linear models explaining best ways their results.