Abstract
A new multiple linear regression model has been proposed based on the synchronization of the raw discrete data and the estimated set of dead-times. In this method, intermittent discrete variables would be adopted, if necessary, in addition to the ordinary synchronous discrete variables with the dependent variable. The observational interval consistency of these intermittent data is obtained by natural spline interpolation and resampling. The optimal dead-times are included in the model with respect to explanatory variables. This estimation of dead-time set is performed by the coefficient of determination and t-test. The proposed model has been applied to estimate a price formation function for industrial natural rubber. The goodness of fit of this estimated model has been improved as much as 66 percent, which is four times better than that of the ordinary model. This improved model is fairly compatible with persuasive explanation in economics.
