People will obtain many relating data of two or more than two dimension during experiments and production. These data will help them to solve problems of reality on contrary, which need data processing to make them become mathematical model reflecting the data variation regulation. The application of the Least Square Method can only make linear regression, but to the nonlinear problems it must construct relating mathematical relationship expression, namely mechanism model through procedure supposing to do linearization processing of mechanism model and then do regression modeling computation. Some relating data of the recursive models are good, but the data of reality are changeable, some deduce mechanism models. After the linear process the correlation property of the regression model is not good, and some relating data even can't deduce in the mechanism model. It is even more harder to build mathematicalematical models. Least Cubic Method solves problems that Least Square Method Data Regression met in the regression of relating data. Since the computers are widely used and applied in experiment, designing and production, it makes the regression computation based on the theory of least Cubic method into reality. People can not only process the mechanism model through the regression linearization processing better, but can also give a sound mathematical model to the relating data which can't deduce a mechanism models.