hossein-f / celestial-classification Goto Github PK

Classification of the celestial objects (stars, galaxis and quasars) is an important issue in astrophysics. It provides a better undrestanding of the universe and its content. In particuar, using bright quasars as reference points, we can create the most detailed three-dimensional map of the universe. I utilise the DR12 (the final data release of the SDSS-III) release of the Sloan Digital Sky Survey (SDSS) for this purpose. I use both spectroscopic and photometric data from DR12 database.

For a review of the phsyics of the problem and details of the database I used, refere to follwoing website or similar ones

https://www.quora.com/What-is-the-difference-between-a-quasar-and-a-star http://skyserver.sdss.org/dr7/en/help/docs/QS_UserGuide.asp http://www.sdss.org/dr12/algorithms/magnitudes/ http://www.astroml.org/examples/datasets/plot_sdss_filters.html

This work is more like a comparison among different algorithms used for the classification of celestial objects in a pedagogical manner for the purpose of education thoses physics and data science students and resaerchers. We start with some data munging and features slection . We then apply the following algorothms:

For each model, we also evaluate the classfiers. The result can be found in the algorithm_comparison file.