CLASSIFICATION OF WHITE DWARF SUBCLASSES USING THE RANDOM FOREST METHOD

• Main Author: Dian Islamiati, Anneke
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/75194
• Institution: Institut Teknologi Bandung

White dwarf stars are the final stage in the evolution of most stars. It is estimated that approximately 97% of all stars will passively end their life cycles by shedding their outer layers and transforming into white dwarfs. Studies on white dwarfs provide information about the evolution of stars from beginning to end. Furthermore, by studying white dwarfs, we can gather information about the chemical evolution of our galaxy. Not only that, but the process of white dwarf evolution also offers insights into the properties of degenerate matter. In this Final Project, classification will be conducted to determine the subclasses of white dwarfs and the regression of their physical parameters, namely effective temperature and surface gravity. The method used is Random Forest, which is a machine learning method built from multiple decision trees to perform its task. To obtain the values of the physical parameters and classify the subclasses of white dwarfs, this Final Project will utilize the spectra of white dwarfs with wavelength as the feature. The spectra of white dwarfs used in this Final Project were obtained from Gaia DR3 and LAMOST DR8 data. In the classification phase, this Final Project will classify two subclasses of white dwarfs, namely the DA and DAZ subclasses. The results of this classification will then be compared with the paper Echeverry, D., dkk. (2022), which distinguishes objects between white dwarfs, main sequence stars of class M, and main sequence-white dwarf binaries. Meanwhile, the results of the regression will be compared with the error values obtained from the observation instrument and available in the database. The accuracy for the classification process reaches 90%, while the error values for the physical parameters are better compared to the errors in the instrument.