Data Mining in Astronomical Survey Data of Variable Stars with Python
summary
A reproducible Python pipeline processes the OGLE-III catalog of ~400,000 variable stars, extracting light-curve features using numpy/scipy, astropy, astroML and scikit-learn, computing periods with Lomb–Scargle–style methods and Fourier components, and then clustering and classifying the objects. Period estimates agree with OGLE published values to about 10^-3 days, unsupervised and supervised labels match at roughly 96–99% depending on method, the implementation runs on a desktop (random forest training ~45 minutes), is open-source, and is validated end-to-end against OGLE.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 5:26Scientific goals
- 10:43Python in astronomy
- 12:07Lightcurve analysis
- 20:10Data mining stats
- 22:14ML classification
- 32:16Long-term variability outlook
- 35:29Conclusion
- 36:34Q&A
related talks
37mCounting the Stars
28mNot a long time ago, in a galaxy not very far away, an astronomer and a computer scientist walk into
36mTwinkle twinkle little star, how I wonder what you are
45mStars, Planets, and Python
43mIs There A Needle In The Haystack? Using Python To Narrow Down The Search For Life I
22mDiscovering the mysteries of the universe with Python
23mUsing a Bayesian Neural Network in the Detection of Exoplanets
30mIrregular time series and how to whip them