IceCube – Neutrinos in Deep Ice (Kaggle competition data)

IceCube Kaggle challenge (2023): simulated events (pulses per sensor) with true directions, for ML reconstruction

Data sets IceCube Atmospheric Astrophysical (high energy) Detector response Statistical methods Curated

Data of the public Kaggle competition "IceCube – Neutrinos in Deep Ice" (2023), organised by the IceCube Collaboration: simulated neutrino events as lists of pulses (sensor, time, charge) with the sensor geometry and the true neutrino direction, for the task of reconstructing the arrival direction with machine-learning methods. One of the largest public neutrino-telescope simulation samples and a standard benchmark for ML reconstruction.

Data card 4/11 sections

Summary

Large simulated IceCube event sample with ground-truth directions, released as an ML challenge.

Intended use

Developing and benchmarking machine-learning event reconstruction for neutrino telescopes.

Out-of-scope use

Physics analyses: the events are simulated, not detector data, and come without the full systematics treatment.

Known limitations

Access requires a Kaggle account and acceptance of the competition rules; no DOI.

FAIRFindable Accessible Interoperable Reusable registration; competition-rules licence

Get it

LocationRoleHostFormatsSize
Kaggle competition data primaryOtherParquet

Metadata

CreatorsIceCube Collaboration
PublisherKaggle
Published2023
LicenceProprietary / custom terms
Kaggle competition data: use under the competition rules (Kaggle account and acceptance of the rules required)
AccessRegistration
Experiments IceCube
Detector technologyIce Cherenkov
Data levelsMonte Carlo
Observablespulse sensor id, time, charge, true direction
Covariance / systematics / MC / / yes

Used by

Know of another paper or code that uses this, or an input it depends on? Tell us.

Papers

Related

Curation

StatusCurated (candidate: harvested, not reviewed · curated: reviewed metadata · verified: resource downloaded and checked against its paper)
Added2026-10-23
Curator notes
• sample size and file layout not recorded here (Kaggle data page needs a login)

Edit this record · Report a problem · JSON