A space-structure based dissimilarity measure for categorical data

The development of analysis methods for categorical data begun in 90's decade, and it has been booming the last years. On other hand, performance many these depends on used metric. Therefore, determining a dissimilarity measure is one most attractive recent challenges mining problems. However, several similarity/dissimilarity measures proposed literature have drawbacks due to high computational cost, or poor performance. For this reason, we propose new distance metric data. We call it: Weighted pairing (W-P) based feature space-structure, where weights are understood like degree contribution an attribute compact cluster structure. W-P was evaluated unsupervised learning framework terms quality index. test six real datasets downloaded from public UCI repository, make comparison with (DM3) method hamming (H-SBI). Results show that our proposal outperforms DM3 H-SBI different experimental configurations. Also, achieves highest rand index values better clustering discriminant than methods.