MagicTelescope: Subsampled Particle Physics Classification Data

MagicTelescope_seed_2_nrows_2000_nclasses_10_ncols_100_stratify_True is a subsampled version of the original MagicTelescope dataset from OpenML. The dataset was created by Eddie Bergman using a controlled random sampling process to select 2000 rows, 100 columns, and 10 classes. It is licensed as public domain in the United States (us-pd).

Use Cases

• Benchmarking classification algorithms on a controlled subset of particle physics data.
• Training models to distinguish between signal and background events in telescope observations.
• Analyzing feature importance within a reduced set of 100 simulated variables.
• Testing model performance on a stratified sample containing 10 distinct event classes.

Strengths

• Subsampling parameters are explicitly documented: seed=2, nrows=2000, ncols=100, nclasses=10.
• The creation process used stratified sampling to preserve class distribution, as indicated by the code.
• Derived from a known source dataset (MagicTelescope ID 44125) on OpenML.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown for the original source, and the total size of this subsample is unspecified.
• Last update date is unknown; freshness unverified.

Provenance

Source

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

Collection Method

Algorithmic subsampling via a Python script using random seed and stratified train_test_split.

Time Range

null

Freshness

null

Geography

null