---
language:
- en
tags:
- adult
- tabular_classification
- binary_classification
pretty_name: Adult
size_categories:
- 10K<n<100K
task_categories:  # Full list at https://github.com/huggingface/hub-docs/blob/main/js/src/lib/interfaces/Types.ts
- tabular-classification
configs:
- encoding
- income
- income-no race
- race
---
# Adult
The [Adult dataset](https://archive.ics.uci.edu/ml/datasets/Adult) from the [UCI ML repository](https://archive.ics.uci.edu/ml/datasets).
Census dataset including income threshold.

# Configurations and tasks
The dataset has four configurations:
- `encoding`, which holds the encoding dictionaries mapping binary and ordinal features to their value;
- `income`, for binary classification of the individual's income;
- `income-no race`, as `income`, but the `race` feature is removed;
- `race`, multiclass classification to predict the `race` of the individual.

# Features
- `age` Age of the person;
- `capital_gain` Capital gained by the person;
- `capital_loss` Capital lost by the person;
- `education` Education level: the higher, the more educated the person;
- `final_weight`
- `hours_per_week` Hours worked per week;
- `marital_status` Marital status of the person;
- `native_country` Native country of the person;
- `occupation` Job of the person;
- `race` Race of the person;
- `relationship`
- `sex` Sex of the person;
- `workclass` Type of job of the person;
- `over_threshold` `1` for income `>= 50k$`, `0` otherwise.