Gretel.ai: US Census Income Dataset (UCI) – Reduced Bias

Predict whether income exceeds $50k/yr. Synthetic data available to balance race, income bracket, and gender.

Description:

This dataset is based on the popular “Adult Data Set” or “Census Income” dataset published by the University of California Irvine ML repository. It is commonly used to predict whether income exceeds $50k/yr based on census data.

In this dataset, synthetic data is used to boost under-represented `race`, `gender`, and `income_bracket` classes from the US Census dataset using a synthetic data model trained on the original census data. The goals for this dataset are to improve representation bias, where algorithms trained on the task would inherently favor groups with greater representation to create a more fair, and less biased dataset. This dataset may also be used to support research efforts around AI ethics and bias.

Tables Included:
– US Census data (original)
– US Census data (synthetic) – a complete synthetic dataset with representation bias removed

Original dataset: http://archive.ics.uci.edu/ml/datasets/Adult

Fields/Attributes Included:
– Age
– Workclass
– Income-bracket
– Education
– Education-num
– Marital-status
– Occupation
– Relationship
– Race
– Sex
– Capital-gain
– Capital-loss
– Hours-per-week
– Native-country

Example Use Case:
Create a prediction task to determine whether a person makes over $50k a year.

About the Provider:

Gretel.ai is an advanced synthetic data platform featuring simple APIs and an open-source AI-based core. Developers and companies use our synthetic data APIs to remove biases in their machine learning datasets, and to enable access to synthetic versions of sensitive data featuring mathematical guarantees around privacy.

Visit the provider’s website for more information