Oral Abstract

Invited talk (I4.1) Daniela Huppenkothen ()

Theme: Data science challenges: tools from statistics to machine learning

Data Science Challenges in Time Domain Astronomy: Building Methods, Tools and Communities

Light-curves — time series data of the brightness of astronomical sources— have always been integral to understanding the physics governing many of the phenomena we observe. Recent and future facilities such as LOFAR, Kepler, TESS, ZTF, NICER, NuSTAR, LSST and SKA provide a range of new capabilities to probe the night sky at different timescales and wavelengths. However, they also generate dramatically larger and more complex datasets than their predecessors, which in turn require new methods, technologies and tools to analyze them effectively and efficiently. In this talk, I will present examples of particular challenges
in astronomical time series analysis, and demonstrate how modern data-science
methods can help us uncover new insights about the universe.
However, simply adopting new methods is not enough. Many of the most significant challenges are social, rather than simply technical: how do we enable collaboration across fields? How do we cut through discipline-specific jargon? How do we build data science communities that are innovative and inclusive? I will discuss some of the community-building aspects of this emerging field of research, and show how community building serves as a pre-requisite for, and driver of, scientific innovation.