(pandas, numpy, tensorflow/keras, scikit-learn & matplotlib)
Note: There are 2 notebooks photometry.ipynb (models were initially created and ensembled here) and production.ipynb (already saved models were loaded and used in production)
Large catalogs of unlabelled stellar objects are available. Labelling these stellar objects is important for a number of reasons. E.g: for statistical poulation analyses and for testing cosmological models to name a few. Although these stellar objects can be classified by analyzing their optical spectrums but that process is time consuming. Next generation of telescopes will increase the quantity of available unlabelled data even more! That's why I have tried to use the photometric data and a combination of machine learning approaches to label the stellar objects.
Table of Contents photometry.ipynb
The results were:
- For objects correctly_classified as Quasars, 90.3% of them had a probability greater than 0.9 of being a Quasar.
- For objects correctly_classified as Galaxies, 86.7% of them had a probability greater than 0.9 of being a Galaxy.
- For objects correctly_classified as Stars, 99.6% of them had a probability greater than 0.9 of being a Star.
Screenshots from the notebook production.ipynb:
Feature Name | Description |
---|---|
u | Ultraviolet filter in the photometric system |
g | Green filter in the photometric system |
r | Red filter in the photometric system |
i | Near Infrared filter in the photometric system |
z | Infrared filter in the photometric system |
redshift | Redshift value based on the increase in wavelength |
class | Object class (galaxy, star, or quasar object) |