When one thinks of Astronomy, they picture stars, moons, planets, and galaxies beyond galaxies – sometimes, even complex mathematical symbols and unending numerical equations pop into the imagination. But beyond this, unless one is already deep entrenched in the fantasy of outer space, it’s seldom that the idea of Data Driven Astronomy (DDA) comes into play.
But looking at Astronomy through the lens of Data Science, I’m sure that even as you read – ideas have begun to explode into your minds. Classification – one of the primary aspects that we can think of even without knowing the depths of Astronomy – may have been the first idea you thought of. If we look into denser concepts, we may even find a niche for regression in the field. And this is just the basics that most that know ML would think of.
Looking into the vast amount (measuring in terabytes, petabytes and even exabytes) and variety (pictures, feeds, time series) of data that is generated in the field of Astronomy, we’d consider it ridiculous to think of people pouring over stack-full of numbers and figures on a continuous basis. There must be some automated computer programs working through the generated data in order to analyse the data in most efficient manner. And these algorithms? They’re all a part of the larger field of DDA.
If we look into the data generation in Astronomy, we’d find that the generated data already qualifies for the tag of “Big Data” with its Velocity, Volume, Variety, Veracity, and the Value that it can be utilized to generate.
There are also examples and case studies where data mining procedures like classification, regression, clustering, outlier-detection, time series analysis as well as feature selection/dimension reduction have been applied to the Astronomical data.
So… where and how can one interested in the field learn more about it? Following are the few resources we found interesting as an introduction to interested explorers:
A Proceedings Paper by Yanxia Zhang and Yongheng Zhao, it explains in detail various aspects of the Data Science approach towards Astronomy. It goes over the basic definitions of related fields to astronomy, namely – Astrophysics, Astro-statistics and Astro-informatics, and touches upon how various data generated in the field qualifies as “Big-Data.” We also learn of some standard data mining procedures that have been used in the field, in order to understand the feasibility of Data Driven Astronomy. The paper ends with giving references to various tools that can be used for DDA as well as some organizations and conferences that can be referred to by interested students.
A collaborative project aimed at classifying galaxies on the basis of inputs given by contributors around the globe – Galaxy Zoo was launched in 2007 and its first phase ended in 2009, but enthusiasts and explorers around the world still go to the site and help in manual classification of galaxies based on the questions/parameters specified for subsequent phases of the project. While the working of this project doesn’t actively use Machine Learning in the traditional sense, we do get an understanding of the basics criteria and parameters of galactic classification – not to mention that their catalogue full of public domain datasets are nothing to sniff at for those who are interested to practice with their own classification models.
Offered by the University of Sydney on Coursera, by Associate Professor Tara Murphy and Post-doctorate Researcher Simon Murphy, this certification offers enthusiasts and students of Astronomy, a comprehensive introduction to DDA through a 6-week module that touches upon basics like querying through SQL to relatively advanced concepts of regression and random forest classification, along with a basic heads-up on managing relatively high quantity of data. The coding involved is in Python for this course.
A set of 26 videos of about 16 hours that comprise of multiple aspects of DDA, covering basics of a theoretical introduction as well as examples of practical application of DDA, this playlist is an interesting route of exploration for those who have already gone over the resources mentioned before. Someone unfamiliar to the field may need to refer to other resources to understand everything in the playlist, but that would also add to the learning path, setting up a stronger foundation for anyone who desires to embark on further exploration in the field of DDA and Astronomy.
We believe that these resources would act as a starter guide to all who are interested before they step into the second phase of their learning and try on hands-on projects using the multitude of available datasets that are available on the web. So, for all the new explorers embarking in their quest of galaxies beyond – Happy Learning!
By-
Katyaini Ranjan Choudhary
Comments