A Flexible Grey Incidence Degree and Application for Unequal Length Sequence

摘要

Traditional Grey incidence degree (GID) methods usually align unequal sequences by data removal, mean statistic or prediction and so on. These manual interventions have a great impact on the performance of GID methods. Aiming at reducing the impact of manual interventions, a novel GID based on Dynamic Time Warping under limited warping path length (GID-LDTW) is proposed. It originates from the classical Deng's GID, and measures the distance between sequences with LDTW path. First, DTW distance is introduced to describe the incidence between sequences by bending the time axis. Next, the optimization limits for the total number of links are used to automatically decide how many links to allocate between data points and where to place these links. Finally, LDTW distance is rectified by an average processing to replace absolute difference with strict matching. The analysis of case shows GID-LDTW is more accurate and reasonable. To observe the actual application effect of GID-LDTW, grey incidence cluster (GIC-LDTW) is conducted. Comparative experiments on five different datasets show that the proposed method is superior to GIC methods based on traditional GID models. Especially, when the length of the sequence is inconsistent, the clustering result of this new method is better. These fully show that GID-LDTW is more effective and reasonable for unequal length sequences.