Electricity: Subsampled Classification Dataset for Machine Learning Benchmarking

A 2000-row, 100-column, 10-class subsample of the original electricity dataset, generated with a specific random seed. The dataset was created by Eddie Bergman using a Python script for uniform sampling of rows, columns, and classes. Its license is us-pd, indicating it is in the public domain in the United States.

Use Cases

• Benchmarking classification model performance on a controlled subset of electricity data.
• Comparing feature selection methods on a dataset with 100 sampled columns.
• Testing model scalability and training time on a dataset of 2000 rows.
• Evaluating stratified sampling techniques for maintaining class distribution.

Strengths

• Controlled subsample size of 2000 rows and 100 columns, enabling consistent benchmarking.
• Stratified sampling was used to preserve the original dataset's class distribution.
• The generation process is fully reproducible due to the specified random seed of 4.

Limitations

• The original dataset's column names and specific features are unknown, limiting interpretability.
• Row count and file size are unknown for this specific subsample.
• The description lacks details on the temporal coverage and geographic scope of the underlying data.

Provenance

Source

Subsampled from the 'electricity' dataset (ID 44120) on OpenML.

Collection Method

Generated via a Python script performing random uniform sampling of rows, columns, and classes.

Time Range

null

Freshness

Last updated date is unknown; freshness unverified.

Geography

null