Skip to content. | Skip to navigation

Personal tools


You are here: Home / Organizations / AAS Working Group in Astroinformatics and Astrostatistics / Astroinformatics and Astrostatistics in Astronomical Research: Steps Towards Better Curricula

Astroinformatics and Astrostatistics in Astronomical Research: Steps Towards Better Curricula

At the 225th American Astronomical Society meeting in Seattle, January 2015, the Working Group organized a session to discuss educational and professional implications of the rise of computational and statistical needs in the astronomical community. Notes from the session follow below, written by WGAA Chair Aneta Siemiginowska.

WGAA organized a Session on Astroinformatics and Astrostatistics in Astronomical Research Steps Towards Better Curricula at the 225th AAS meeting in Seattle. The session was scheduled for January 7, 2015 and was very well attended with the standing room only. The session participants included students, postdocs and faculty.

The purpose of the session was to highlight the importance of data analytics training in astronomy, both for the sake of astronomical research and in order to make astronomy graduates more employable. Although astronomy and astrophysics are witnessing dramatic increases in data volume as detectors, telescopes, and computers become ever more powerful, the traditional training of astronomy and physics students is not providing skills to handle such voluminous and complex data sets. Equally worrisome, research funds and hiring options in astronomy are diminishing; in particular, a number of candidates for permanent (or steady) jobs significantly exceeds the job availability. As a result many of astronomy graduates have transitioned out of astronomy to work in areas where their analytic skills become highly valuable. The main goals of the proposed session were to discuss ways to improve Big Data training and research in astronomy, as well as to explore the connections between data science in astronomy and in the other research or technology areas where astronomy postdocs or recent graduates could excel and compete.

We invited two speakers, recent astronomy Ph.D. graduates, who recently transitioned to tech industry to critically compare academic and industrial environments. Arfon M. Smith who received the PhD in Astrochemistry and currently works as a Scientist at GitHub Inc. talked on “Working on interesting problems” describing his transition steps, motivation and aspects of his current work. The slides from his talk and the blog entry are available on the web page his web pages, and

He pointed out that usually a code written in support of the research is not well credited in astronomy, even if it is widely used by the community. However, publishing a paper describing the code is not the most effective/efficient process in obtaining the credit, and the number of citation of such published paper is usually much lower than the actual use of the code (see Astropy paper which has been cited 53 in contrast to about thousands downloads of the code from the repository). Sharing the code, licensing it and making it available in code databases and repositories such as GitHub, is recognized by the tech industry.

The second speaker, Jessica Kirkpatrick, received her PhD in Astrophysics from Berkeley in 2012. After an exhaustive job search within academia and beyond, she accepted a job as a data scientist/analyst for the social network Yammer (acquired by Microsoft) and is now the Director of Data Science for Education Company InstaEDU. Now instead of spending her days finding patterns in the large scale structure of galaxies, she finds patterns in the behaviors of people. She talked about her transition from astrophysics to tech, compared and contrasted the two fields, and gave tips about how to land a tech job. She also discussed useful tools which helped her with her transition. Her slides are available at

Jessica points out that there are more data analysis positions outside academia and the data scientist work is often more collaborative than the postdoc one, with faster pace and more variety. She advices on working on side projects, learning Python and SQL, getting on LinkedIn and staying networked. 

The panel discussion followed these two presentations. The panel members included George Djorgovski (Caltech), Eric Feigelson (PennState), Zeljko Ivezic (U of Washington), Jessica Kirkpatrick (InstaEDU), Arfon Smith (GitHub Inc), and discussion was moderated by Aneta Siemiginowska (Center for Astrophysics). 

Some main discussion points and issues during the session included:

  1. Training faculty is important. Weekly courses for faculty during the teaching breaks could help. There are also courses available online that should be advertised. Advice on which course is worth taking would be helpful.
  2. Funding the courses for faculty is needed. Suggestion of a proposal to NSF to get funding for one/two weeklong courses. Logistics of such training and organization is also needed.
  3. The community should encourage the Data Science training of the students and faculty.
  4. The concerned was raised about advisors who may not support training in data science. It is understandable that the students may have difficulties in learning if they do not find the support in their immediate advisors. The community needs to be aware of such situations.
  5. The changes to the curriculum are often impossible, or take a long time to implement. Often an addition of a new course means a deletion of another course. Such modifications to the curriculum need to be carefully considered and require a strong motivation for revisions. Any advice on how it is done in other places could help.
  6. There is some room when considering the curriculum change in the required vs optional courses. Optional course if listed can accrete students if it is well designed, interesting and advertised.
  7. We are not alone in this predicament all fields have these same problems. We should work with like-minded faculty in other fields and with the top level administrators to make these changes universitywide. 
  8. Advice to people who want to look for a job outside academia was also given: use Python (not IDL), be active on GitHub, sign for Linkedin account and work on the resume.
  9. Brain drain is happening and is unavoidable. Keeping talented scientists in academia is and will be even more challenging. In the era of the Big Data the astronomy community have to be able to compete with the industry and recognize that the Data Science is an important part of the modern astronomy research.


Aneta Siemiginowska

AAS WGAA Working Group on Astroinformatics and Astrostatistics