MagicTelescope: Simulated Particle Detection Data for Classification

A 2000-row, 100-column subsample of the MagicTelescope dataset, generated with a random seed of 3 and retaining 10 classes. The original dataset, created by Eddie Bergman, contains simulated data from a ground-based gamma-ray telescope for particle detection. The subsample was created using a stratified random sampling method to preserve class distribution.

Use Cases

• Training binary or multi-class classifiers based on simulated particle detection features.
• Benchmarking model performance on a controlled, stratified subset of physics data.
• Feature selection experiments using the 100 sampled columns from the original dataset.

Strengths

• Contains 2000 rows, providing a substantial sample for model training and testing.
• Includes 100 features (columns), offering a multi-dimensional representation of particle events.
• Sampling was stratified, which likely preserves the original class distribution for the 10 retained classes.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count of the original dataset is unknown, which may limit suitability assessment of the subsample.
• Last update date is unknown; freshness unverified.

Provenance

Source

openml, derived from the MagicTelescope dataset (ID 44125).

Collection Method

Subsampled via a Python script using random selection of rows and columns with stratification.