astroML has been nominated for the 2026 World AI Awards in the Aerospace category, recognising an open-source project built to bring machine learning, statistical analysis and data mining into astronomy and astrophysics.
Unlike an aerospace manufacturer building autonomous aircraft or spacecraft, astroML works much closer to the data. The Python library gives astronomers and researchers tools for extracting patterns from complex astronomical datasets — a problem that has grown steadily more important as telescopes and sky surveys generate larger volumes of observations.
The World AI Awards recognises organisations, individuals, products and technologies contributing to the development and real-world application of artificial intelligence across industries.
For astroML, the connection is straightforward. Modern astronomy does not suffer from a shortage of observations. The harder problem is making sense of them.
Machine learning built specifically for astronomy
The astroML project was started in 2012 by Jake VanderPlas alongside the book Statistics, Data Mining, and Machine Learning in Astronomy, written by Željko Ivezić, Andrew Connolly, Jacob VanderPlas and Alex Gray.
It is built around Python’s scientific computing ecosystem, including NumPy, SciPy, scikit-learn, Matplotlib and Astropy. Rather than trying to create another general-purpose AI framework, astroML brings together statistical and machine-learning methods particularly useful for astronomical research.
Its toolkit covers classification, regression, clustering, density estimation, dimensionality reduction, correlation functions and time-series analysis. It also provides loaders for open astronomical datasets and worked examples showing how those techniques can be applied.
That makes the project useful in a field where researchers regularly confront huge catalogues containing measurements of stars, galaxies and other astronomical objects.
Turning observations into something researchers can analyse
The practical side of astroML becomes clearer in its examples.
One demonstrates the use of random forest regression to estimate photometric redshifts from Sloan Digital Sky Survey data. Redshift helps astronomers understand the distance and movement of astronomical objects, while photometric approaches can estimate it from imaging measurements when spectroscopy is unavailable or impractical at scale.
Other astroML resources explore clustering, density estimation, regression and astronomical time series.
There is also a deep-learning example for classifying astronomical images, using TensorFlow and Keras to move from a basic neural network toward convolutional neural-network architectures.
These are not claims that astroML itself has discovered a new galaxy or operates an observatory. It is infrastructure for researchers: reusable code, datasets and examples that make sophisticated analytical techniques easier to apply to scientific questions.
Astronomy has a data problem — and that matters
astroML emerged from a problem that has only become more relevant.
The research paper introducing the project described astronomy as entering an era of rapidly expanding datasets as increasingly capable telescopes, detectors and sky surveys produce measurements for enormous numbers of astronomical sources. The authors positioned astroML as a practical collection of machine-learning tools for researchers and students dealing with that complexity.
That is where machine learning becomes useful without needing the usual AI theatrics.
Astronomers need to classify objects, estimate properties from incomplete observations, detect structure, identify unusual behaviour and analyse signals that change over time. Many of those jobs involve finding meaningful patterns inside datasets too large or multidimensional for straightforward manual analysis.
astroML packages those techniques into an openly available environment researchers can inspect, modify and reproduce.
Its accompanying interactive book pushes the idea further. Browser-based notebooks connect statistical and machine-learning concepts with executable examples using real astronomical datasets, covering subjects from Bayesian methods and dimensionality reduction to classification and time-series analysis.
Graham Cooke, President of the World AI Awards, said:
“astroML’s nomination highlights an important side of artificial intelligence that can sometimes be overlooked: the tools researchers use to turn enormous scientific datasets into knowledge.
“By making machine learning and statistical methods more accessible for astronomy and astrophysics, astroML shows how open-source technology can support researchers working with increasingly complex observations of the universe. We congratulate the astroML community on its 2026 World AI Awards nomination and look forward to seeing how these tools continue to develop.”
astroML joins organisations, researchers, open-source communities and technology developers being recognised through the 2026 World AI Awards.
The programme recognises organisations, individuals and technologies contributing to the development and application of artificial intelligence across industries.
astroML occupies an interesting place in that landscape. It is not trying to automate a spacecraft or build a commercial AI service. Instead, it provides researchers and students with practical tools for applying machine learning to astronomy — from classifying observations and estimating astronomical properties to analysing time-dependent signals.
As astronomical datasets continue to grow, that quieter layer of scientific infrastructure matters. Better instruments can collect more information about the universe. Researchers still need the statistical and computational tools to understand what the data is telling them.
Learn more about astroML, its documentation and machine-learning tools for astronomy at https://www.astroml.org/.
Discover the World AI Awards 2026, explore the nominees and learn more about the awards at https://www.worldawards.ai/.
