Threshold regression is a method in time series and dynamic data processing, based on thresholds variables and threshold values. Threshold regression was used to improve existing single curve method. The case studies show that the new method has higher accuracy and better stability than the existing metho.