ATLAS-Higgs-Boson-Machine-Learning-Challenge-2014: Higgs Boson Classification Data

The 2014 Kaggle Higgs Boson Machine Learning Challenge dataset contains 800,000 simulated particle collision events from CERN. It was originally sourced from CERN's open data portal and prepared for a public machine learning competition. The data is provided under a CC0 1.0 public domain license.

Use Cases

• Training binary classification models to distinguish Higgs boson signals from background noise based on simulated particle collision features.
• Benchmarking machine learning algorithms on a well-known competition dataset from the physics domain.
• Studying feature engineering and model performance for imbalanced classification tasks in scientific simulations.

Strengths

• Contains 800,000 samples, which is a substantial size for model training.
• Data is sourced from CERN, a leading authority in particle physics research.
• The dataset is a canonical benchmark from a major public machine learning competition.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is known, but the specific features and their distributions are not described.
• Last update date is unknown; freshness unverified.

Provenance

Source

CERN Open Data portal (record 328), prepared for the Kaggle Higgs Boson Machine Learning Challenge 2014.

Collection Method

Simulated particle collision data from the ATLAS experiment.

Time Range

Associated with the 2014 machine learning challenge.