IWDMS2018 - International Workshop on Dark Matter and Stars

Machine learning technique for morphological classification of galaxies

Not scheduled

30m

Anfiteatro Abreu Faro (Center for Astrophysics and Gravitation, Instituto Superior Técnico, University of Lisbon)

Poster session

Speaker

Dr Daria Dobrycheeva (Main Astronomical Observatory of the NAS of Ukraine)

Description

We checked classifiers as Naive Bayes, Random Forest, and Support Vector Classifier on sample of galaxies from SDSS DR9 (N=60561, 0.02<z<0.06). We used the absolute magnitudes Mu, Mg, Mr, Mi, and Mz, all the color indices, and inverse concentration indexes R50/R90 to the center as the attributes of galaxy. To define an accuracy of the mentioned above classifiers we applied the 5-folds validation technique. It turned out that the Random Forest method provides the highest accuracy, namely 91 % of galaxies from the sample were correctly classified (96 % for E and 80 % for L types). The accuracy of other classifiers was from 85 % to 90 %. We were able to classify 60561 galaxies from the SDSS DR9 with unknown morphologies with a good accuracy onto two clasees (47 % E and 53 % L types of galaxies). Finaly, we found 28 199 E and 32 362 L types among them. We able to classify low-redshift galaxies from the SDSS with unknown morphologies with a good accuracy.