Classifying Y-Short Tandem Repeat Data: A Decision Tree Approach

Abstract

Classifying Y-Short Tandem Repeat data has recently been introduced in supervised and unsupervised classifications. This study continues the efforts in classifying YSTR data based on four decision tree models: CHisquared Automatic Interaction Detection (CHAID), Classification and Regression Tree (CART), Quick, Unbiased, Efficient Statistical Tree (QUEST) and C5. A data mining tool, called IBM Statistical Package for the Science Social Modeler 15.0 (IBM® SPSS® Modeler 15) was used for evaluating the performances of the models over six Y-STR data. Overall results showed that the decision tree models were able to classify all six Y-STR data significantly. Among the four models, C5 is the most consistent modelm where it had produced the highest accuracy score of 91.85%, sensitivity score of 93.69% and specificity score of 96.32%.