IMPRS Summer School: Astrostatistics & Data Mmining

IMPRS Summer School: Astrostatistics & Data Mmining IMPRS Summer School: Astrostatistics & Data Mmining
  • Contact

    International Max Planck Research School for Astronomy and Cosmic Physics at the University of Heidelberg (IMPRS-HD)

  • Working language

The information is outdated?

Please let us know

The summer school is open to all graduate students and junior postdocs working in astronomy anywhere in the world.

Lecturers:
Dan Foreman-Mackey (University of Washington)
Shirley Ho (Carnegie Mellon University)
Daniela Huppenkothen (New York University)
Robert Lupton (Princeton University)

Organisation:
IMPRS for Astronomy and Cosmic Physics at the University of Heidelberg:
Max Planck Institute for Astronomy, Max Planck Institute for Nuclear Physics, Astronomisches Rechen-Institut, Institute for Theoretical Astrophysics, Landessternwarte Koenigstuhl, and Heidelberg Institute for Theoretical Studies.
Scientific organizing committee: Coryn Bailer-Jones (MPIA, Heidelberg).

Scope of the School:
Astronomical data sets are increasing in size and complexity. Extracting meaningful and reliable scientific results from these is a challenge, and relies on the application of appropriate and properly-understood methodology.

The school will look at the principles of inference and methods of astronomical data analysis and data mining. We will cover a range of numerical and statistical techniques and their application to different types of astronomical data.

The emphasis will be on practical methodology and solving real-world problems to address questions like: How do we deal with missing data, outliers, and selection effects? How should we compare models to data? How do we best combine different data sets? What is an efficient numerical approach? Topics to be to covered include:
- time series analysis, including quasi-periodicity and stochastic processes
- low photon count data
- multi-dimensional interpolation
- image processing: optimal addition and subtraction
- non-Gaussian and correlated noise
- Bayesian model comparison vs. orthodox hypothesis testing
- MCMC methods
- hierarchical models
- parameter estimation in high dimensional problems
- novelty detection
- handling and visualizing large or multi-dimensional data sets

School format:
The school has four main components spread throughout the week
1. A series of structured lectures given by the four lecturers.
2. Computer-based exercises based on the topics given in the lectures. Some lecturers will integrate the exercise sessions into their lectures.
3. Presentations by other speakers on the application of astrostatistics to open scientific problems.
4. A social programme to enable and encourage scientific interaction between students, lecturers and speakers.

All students will need to bring a laptop with the necessary (free) software installed in advance. Details will be circulated in due course.

More info on this website!

Track this event on your google calendar