Adult Income Data for Classification Tasks

A cleaned version of the UCI Adult Income dataset, prepared for machine learning. It contains individual records with demographic and employment attributes used to predict income level. The original data was sourced from the 1994 U.S. Census database.

Use Cases

• Predict income category (e.g., >50K or <=50K) from demographic features like age, education, and occupation.
• Analyze the relationship between workclass, hours-per-week, and the target income label.
• Evaluate model fairness by testing predictions across sensitive attributes such as race and sex.
• Perform feature importance analysis to identify which attributes (e.g., capital-gain, marital-status) most influence income prediction.

Strengths

• Pre-cleaned for machine learning, reducing initial preprocessing effort.
• Based on the widely used UCI Adult dataset, a standard benchmark for classification.

Limitations

• Dataset size and specific row count are unknown.
• Data originates from 1994 U.S. Census and may not reflect current economic or demographic trends.
• Potential class imbalance in the income label is not quantified.

Provenance

Source

UCI Machine Learning Repository (Adult dataset).

Collection Method

Extracted from the 1994 U.S. Census database.

Time Range

1994

Geography

United States