Multilabel Classifier Chains Algorithm Based on Maximum Spanning Tree and Directed Acyclic Graph

The classifier chains algorithm is aimed at solving the multilabel classification problem by composing labels into a randomized label order. effect of this depends heavily on whether order optimal. To obtain better ordering, authors propose based maximum spanning tree and directed acyclic graph. first uses Pearson's correlation coefficient to calculate between constructs labels, then calculates mutual decision difficulty transform graph, it topological ranking output optimized ordering. Finally, use train predict against Experimental comparisons were conducted proposed other related algorithms seven datasets, ranked second in six evaluation metrics, accounting for 76.2% 16.7%, respectively. experimental results demonstrated effectiveness affirmed its contribution exploring utilizing label-related information.