Variables on economic activities are of random and correlation generally, while ordinary least square regression (OLSR) can not reflect such data. The reason is OLSR defines only the dependent variable as the one-dimensional random variable, without considering the random variable, and the regression results are affected by the chosen coordinate. To improve it and reflect real economic variables, a new independent-coordinated linear regression method -- the rotating inertia method is presented, which is based the dynamics nature of rigid plate fixed-axis rotation. The simulation example verified and compared with the least squares method to conventional method regression system deviation is small, the regression accuracy. Moment of inertia method is a numerical solution, the computation of the advantages make it has a broad application prospect.