Dependent function interval parameters training algorithm based on DBSCAN clustering

Dependent function is used to describe basic-element have a nature in what degree in domain, the interval parameters of dependent function decide the boundery value by which element change from quantitative to qualitative. This paper research on cleaning noise data and clustering with DBSCAN algorhithm based on a set of training samples without regard to subjective factors and computing the interval parameter with clustering result. In the paper, we have two simulations on experiment data and actual data taking the case of elementary dependent function, the simulation results are considerably accurate and reasonable.