Improved Technique for Dimensionality Reduction: Star and Quasar Classification with Typical Testors

This work compares the feature selection technique offered by Testor Theory, using the Yablonski and Compatible Sets (YYC) algorithm for the calculation of a fitting typical testor, against the dimensionality reduction given by the principal component analysis (PCA) calculation. Using the results obtained from the previous algorithms and using a Support Vector Machine (SVM) as a classification model, we acquire the classification results of stars and quasars. Lastly, we analyze the advantages shown by typical testors from the experimental results obtained over the ones obtained from the PCA technique.