Ozone Level 8-Hour Subsampled Dataset with 10 Classes and 100 Features

A 2000-row, 100-column subsample of the OpenML ozone-level-8hr dataset, generated with a specific random seed. The subset includes up to 10 target classes and was created using stratified sampling. Eddie Bergman is listed as the author, and the data is released under a US public domain license.

Use Cases

• Train classification models to predict ozone level categories based on atmospheric measurements.
• Benchmark feature selection methods using the 100 available columns.
• Study the impact of stratified sampling on model performance for imbalanced environmental data.
• Compare algorithm performance on a controlled, medium-scale tabular dataset derived from a larger source.

Strengths

• The subsampling process is fully reproducible, with a specified random seed of 0.
• The dataset has a defined structure of up to 2000 rows and 100 columns, as per the generation parameters.
• Stratified sampling was used, which likely preserves the class distribution of the original data.

Limitations

• The exact number of rows and columns in the final artifact is unknown, as the input states 'Rows: unknown' and 'Columns: unknown'.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

openml

Collection Method

Programmatic subsampling from the parent dataset 'ozone-level-8hr' (OpenML ID 1487).