Ozone Level 8-Hour: Subsampled Classification Dataset with 10 Classes

A subsampled version of the ozone-level-8hr dataset for classification tasks. The subset was generated with a random seed of 4, targeting 2000 rows, 100 columns, and 10 classes, using stratified sampling. The original dataset is associated with air quality monitoring and was authored by Eddie Bergman.

Use Cases

• Training classification models to predict ozone levels based on atmospheric measurements.
• Benchmarking model performance on a controlled, stratified subset of environmental data.
• Analyzing feature importance for ozone prediction among a curated set of 100 variables.

Strengths

• Subsampling parameters are explicitly documented: seed=4, target rows=2000, target columns=100, target classes=10.
• Stratified sampling was applied, which likely preserves class distribution from the original dataset.
• The dataset is derived from a known OpenML source (ID 1487) focused on a specific environmental measurement.

Limitations

• Row count is unknown; the actual size may differ from the target 2000 rows.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

OpenML dataset ozone-level-8hr (ID 1487), subsampled by Eddie Bergman.

Collection Method

Algorithmic subsampling with random selection of rows, columns, and classes, using stratified sampling where applicable.