Parameter estimation with inverse problem often suffers from unfavorable conditions in the real world. Useless data and many input parameters make the problem complicated and takes long time for calculation. Data refinement and reformulation of the problem can solve that kind of difficulties. In this research, a method to decompose the problem is suggested. Sensitivity analysis is performed and variable-response pairing matrix is formulated. The matrix is diagonalized with replacement of rows and columns. Sequential estimation is performed according to the hierarchy of the diagonalized matrix. The method is tested with speed reducer example and it shows improvement in calculation efficiency by reducing the estimation time by 75% without loss of accuracy.