Electricity Consumption Data: A 2000-Row, 100-Feature Subsampled Classification Dataset

A 2000-row, 100-feature subsample of a larger electricity dataset, created by Eddie Bergman. The data is structured for classification tasks with up to 10 target classes, generated via a controlled subsampling process with a specified random seed. The original dataset is associated with electricity and load forecasting topics.

Use Cases

• Classifying electricity consumption patterns based on the 100 available features.
• Benchmarking model performance on a controlled, stratified subsample of energy data.
• Training predictive models for load forecasting using tabular feature sets.
• Analyzing feature importance for electricity-related classification tasks.

Strengths

• Controlled subsampling with a specified random seed (seed=0) for reproducibility.
• Dataset size is explicitly defined as 2000 rows and 100 columns.
• Stratified sampling was used to preserve class distribution in the target variable.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is known for the subsample, but the size and details of the original source dataset are not provided.
• Last update date is unknown; freshness unverified.

Provenance

Source

openml, derived from dataset ID 44120 (electricity)

Collection Method

Algorithmic subsampling from a parent dataset using a custom Python function.