An empirical investigation of hierarchical squared-error clustering formulations
Christopher L. Huntley · 2002
The authors present an application of simulated annealing to hierarchical clustering. A formal hierarchical clustering problem formulation is presented, and two different instances for squared-error clustering are examined. These two instances are then compared to Ward's hierarchical method for 32 data sets. The results show that Ward's method outperforms direct formulations for most of the data sets.>