Modeling centre-based hard and soft clustering for Y chromosome short tandem repeats (YSTR) data

This paper models: (1) Y-STR data and; (2) Y-STR hard and soft clustering. The Y-STR models are extended and developed to test on three data sets of Y-STR haplogroup and Y-STR Surname. The results show that the hard clustering models and the soft clustering models have their advantages and disadvantages. The soft k-Means model produces a good clustering accuracy of 99.62% for Y-STR haplogroup data, whereas the hard k-Medoids obtains the highest score of clustering accuracy of 99.90% for Y-STR Surname data. This scenario seems to be both models have an equally chance of improving Y-STR clustering performances.