Economics & Public Policy Public Services & Risk use casesNACE M72Binary ClassificationTarget: income_bracketOpenML #1590 ↗

Adult

Name: Adult Census Income Dataset
Creator: Ronny Kohavi and Barry Becker
License: https://creativecommons.org/licenses/by/4.0/

Benchmark demographic dataset consisting of 48,842 records extracted from the 1994 US Census Bureau database, designed to predict whether an individual's annual income exceeds $50,000.

✨ Try in Carla Install Chrome Extension ↗100% In-Browser WebGPU • Zero Cloud Upload

48,842Records (Rows)

14Predictive Features

6 / 9Numeric / Categorical

0.9%Missing Value Ratio

0.07931 - AUC Error

273.20sWebGPU Mean Split

BUSINESS CONTEXT & OBJECTIVEMarket Research & Socioeconomic Analytics (M72)

Business Objective: Adult

01Business Context

Financial institutions, marketers, and public policy researchers require granular socioeconomic indicators to evaluate creditworthiness, segment consumer demographics, and assess workforce equity across regions without intrusive direct income disclosures.

02Analytical Objective

Predict whether an individual earns more than $50,000 annually based on demographic, occupational, and capital investment characteristics.

03Economic & Decision Impact

Enables precise customer lifetime value estimation, targeted financial product marketing, and automated wealth tiering while avoiding costly underwriting errors and mitigating demographic model bias.

ML Benchmark Narrative

As one of the most established tabular benchmarks in machine learning, the Adult dataset achieves ~87% accuracy with tuned gradient boosted trees (LightGBM/XGBoost). Modern zero-shot tabular foundation models (TabICL / Carla) match or exceed tree-based baselines with zero parameter tuning directly in-context.

Source Origin:UCI Machine Learning Repository / US Census Bureau

Creator:Ronny Kohavi and Barry Becker (1996)

License:Creative Commons Attribution 4.0 International (CC BY 4.0)

Reference:Paper / Repo ↗

FEATURE SPECIFICATION & DATA DICTIONARY

Target & Feature Column Definitions

Exhaustive business definitions, measurement units, roles, and target variables across all 15 columns.

Col	Variable / Feature Name	Role	Type	Unit / Scale	Description & Business Meaning
A	`income_bracket`Income Bracket	TARGET	CAT	—	Annual income tier classification indicating whether total income is <=50K or >50K.
B	`age_years`Age	Feature	NUM	years	Age of the individual in years.
C	`employment_sector`Employment Sector	Feature	CAT	—	General sector and type of employer (e.g., Private, Self-emp-not-inc, Self-emp-inc, Federal-gov, Local-gov, State-gov, Without-pay, Never-worked).
D	`demographic_weight`Final Demographic Weight	Feature	NUM	ratio	Statistical sampling weight assigned by the US Census Bureau representing population control totals across demographic segments within each state.
E	`education_level`Highest Education Level	Feature	CAT	—	Highest level of formal education completed (e.g., Bachelors, Some-college, 11th, HS-grad, Prof-school, Assoc-acdm, Assoc-voc, 9th, 7th-8th, 12th, Masters, 1st-4th, 10th, Doctorate, 5th-6th, Preschool).
F	`education_years`Education Level (Numeric)	Feature	NUM	years	Total completed years/stages of formal education represented on an ordinal numerical scale (1-16).
G	`marital_status`Marital Status	Feature	CAT	—	Marital status of the individual (e.g., Married-civ-spouse, Divorced, Never-married, Separated, Widowed, Married-spouse-absent, Married-AF-spouse).
H	`job_occupation`Occupation	Feature	CAT	—	Primary job role or professional category (e.g., Tech-support, Craft-repair, Other-service, Sales, Exec-managerial, Prof-specialty, Handlers-cleaners, Machine-op-inspct, Adm-clerical, Farming-fishing, Transport-moving, Priv-house-serv, Protective-serv, Armed-Forces).
I	`household_relationship`Household Relationship	Feature	CAT	—	Role of the individual within their household (e.g., Wife, Own-child, Husband, Not-in-family, Other-relative, Unmarried).
J	`ethnicity_race`Race / Ethnicity	Feature	CAT	—	Self-reported racial background (e.g., White, Asian-Pac-Islander, Amer-Indian-Eskimo, Other, Black).
K	`gender`Gender	Feature	CAT	—	Biological sex of the individual (Male, Female).
L	`annual_capital_gain_usd`Capital Gains	Feature	NUM	USD	Recorded annual income gains derived from investments and capital asset sales.
M	`annual_capital_loss_usd`Capital Losses	Feature	NUM	USD	Recorded annual capital losses incurred from investments and asset sales.
N	`weekly_work_hours`Weekly Work Hours	Feature	NUM	hours	Reported average working hours per week.
O	`country_of_origin`Country of Origin	Feature	CAT	—	Country of birth or nationality of the individual.

DATASET PREVIEW

Interactive Data Table

Explore rows, feature values, and target labels for Adult.

Download ParquetHF Viewer ↗

LOCAL EXECUTION TELEMETRY

TabICLv2 WebGPU Benchmark Results

Complete performance metrics on persisted IID or non-IID evaluation splits, executed 100% locally in the browser sandbox.

Mean Split Latency

273.20s9 splits • 8 ensembles/split

Total Evaluation Runtime

2458.80sIncludes warmup & sync

Evaluation Protocol

3×3 SplitsRepeated IID • 9 total

TabICLv2 Score

0.07931 - AUC Error

Evaluation Split	Train Rows	Test Rows	Split Duration?Wall-clock inference time taken for this evaluation split running 100% locally via WebGPU.Learn more →	Accuracy?Proportion of correct test predictions across all classes.Google ML Guide & Details →	ROC-AUC?Area under the ROC curve evaluating ranking capability across all thresholds.Google ML Guide & Details →	F1 Score?Harmonic mean of Precision and Recall, robust against class imbalance.Learn more →	Precision?Proportion of predicted positives that were actually positive.Google ML Guide & Details →	Recall?Proportion of actual positive cases successfully captured.Google ML Guide & Details →
R1 / F1	32,561	16,281	274.34s	86.76%	0.9233	0.6924	0.7800	0.6224
R1 / F2	32,561	16,281	274.10s	86.95%	0.9260	0.7057	0.7669	0.6535
R1 / F3	32,562	16,280	272.09s	86.62%	0.9222	0.6979	0.7586	0.6462
R2 / F1	32,561	16,281	273.47s	87.06%	0.9255	0.7043	0.7770	0.6440
R2 / F2	32,561	16,281	273.45s	86.38%	0.9224	0.6771	0.7821	0.5970
R2 / F3	32,562	16,280	272.00s	83.21%	0.9104	0.6863	0.6206	0.7677
R3 / F1	32,561	16,281	273.66s	86.28%	0.9213	0.6837	0.7625	0.6196
R3 / F2	32,561	16,281	273.55s	86.93%	0.9262	0.6994	0.7776	0.6355
R3 / F3	32,562	16,280	272.14s	86.82%	0.9241	0.6997	0.7694	0.6416
Mean ± Std	—	—	273.20s	86.33% ± 1.13%	0.922 ± 0.005	0.694 ± 0.009	0.755	0.648

UNIFIED BENCHMARK LEADERBOARD

Foundation Models vs Traditional ML Baselines

Side-by-side evaluation of our in-browser Carla engine (WebGPU), open-source tabular foundation models, and OpenML baselines.

Rank	Algorithm / Model	Model Family	Runtime	Predictive Accuracy?Proportion of correct test predictions across cross-validation splits.Google ML Guide & Details →	ROC-AUC?Area under the ROC curve evaluating ranking capability.Learn more →	F-Measure?Harmonic mean of precision and recall.Learn more →	Reference / Repo
#1	Google TabFM v1.0	Tabular Foundation Model	PyTorch/Python	87.60%	0.9323	0.7162	google-research/tabfm ↗
#2	EXAONE Tabular	Tabular Foundation Model	PyTorch/Python	87.58%	0.9320	0.7171	LGAI-Research/EXAONE-Tabular ↗
#3	nanotabicl Vanilla	Tabular Foundation Model	PyTorch/Python	86.83%	0.9237	0.6992	soda-inria/nanotabicl ↗
#4	FilteredClassifier J48	Decision Tree	Python	86.72%	0.8933	0.8615	OpenML #1785495 ↗
#5	FilteredClassifier MultiSearch LogitBoost REPTree	Logistic / Linear Model	Python	86.63%	0.9199	0.8603	OpenML #1838085 ↗
#6	DTNB	Machine Learning Model	Python	86.43%	0.9178	0.8617	OpenML #1810088 ↗
#7	SimpleCart	Machine Learning Model	Python	86.42%	0.8901	0.8573	OpenML #1666057 ↗
#8	TabPFN v3	Tabular Foundation Model	PyTorch/Python	86.38%	0.9201	0.6864	PriorLabs/tabpfn ↗
#9	Carla Engine (TabICLv2 WebGPU)	Tabular Foundation Model	Browser	86.33%	0.9224	0.6941	100% In-Browser
#10	NBTree	Decision Tree	Python	86.32%	0.9078	0.8599	OpenML #1666083 ↗
#11	FilteredClassifier MultilayerPerceptron	Neural Network (MLP)	Python	86.30%	0.8893	0.8566	OpenML #1799965 ↗
#12	Full TabICLv2 (PyTorch Reference)	Tabular Foundation Model	PyTorch/Python	86.23%	0.9207	0.6875	soda-inria/tabicl ↗
#13	Streaming nanotabicl	Tabular Foundation Model	PyTorch/Python	86.18%	0.9178	0.6817	soda-inria/nanotabicl ↗
#14	J48	Decision Tree	Python	86.17%	0.8933	0.8553	OpenML #1687985 ↗
#15	END ND J48	Decision Tree	Python	86.11%	0.8892	0.8552	OpenML #1815759 ↗
#16	A2DE	Machine Learning Model	Python	86.07%	0.9251	0.8628	OpenML #1664564 ↗
#17	Bagging J48	Decision Tree	Python	86.00%	0.9079	0.8547	OpenML #1666377 ↗

STATISTICAL DISTRIBUTIONS

Variable Schema & Summary Distributions

Observed numerical ranges, category cardinalities, missing rates, and sample values across 48,842 rows.

Col	Variable Name	Type	Missing	Distinct	Summary Stats / Distribution	Sample Values
A	`income_bracket`TARGET	CAT	0	2	2 cats: <=50K, >50K	`<=50K<=50K>50K`
B	`age_years`	NUM	0	74	[`17`, `90`] μ=38.6 σ=13.7	`253828`
C	`employment_sector`	CAT	2,799 (5.73%)	8	8 cats: Private, Self-emp-not-inc, Local-gov +5 more	`PrivatePrivateLocal-gov`
D	`demographic_weight`	NUM	0	28,523	[`12285`, `1490400`] μ=189664.1 σ=105602.9	`22680289814336951`
E	`education_level`	CAT	0	16	10 cats: HS-grad, Some-college, Bachelors +7 more	`11thHS-gradAssoc-acdm`
F	`education_years`	NUM	0	16	[`1`, `16`] μ=10.1 σ=2.6	`7912`
G	`marital_status`	CAT	0	7	7 cats: Married-civ-spouse, Never-married, Divorced +4 more	`Never-marriedMarried-civ-spouseMarried-civ-spouse`
H	`job_occupation`	CAT	2,809 (5.75%)	14	10 cats: Prof-specialty, Craft-repair, Exec-managerial +7 more	`Machine-op-inspctFarming-fishingProtective-serv`
I	`household_relationship`	CAT	0	6	6 cats: Husband, Not-in-family, Own-child +3 more	`Own-childHusbandHusband`
J	`ethnicity_race`	CAT	0	5	5 cats: White, Black, Asian-Pac-Islander +2 more	`BlackWhiteWhite`
K	`gender`	CAT	0	2	2 cats: Male, Female	`MaleMaleMale`
L	`annual_capital_gain_usd`	NUM	0	123	[`0`, `99999`] μ=1079.1 σ=7451.9	`000`
M	`annual_capital_loss_usd`	NUM	0	99	[`0`, `4356`] μ=87.5 σ=403.0	`000`
N	`weekly_work_hours`	NUM	0	96	[`1`, `99`] μ=40.4 σ=12.4	`405040`
O	`country_of_origin`	CAT	857 (1.75%)	41	10 cats: United-States, Mexico, Philippines +7 more	`United-StatesUnited-StatesUnited-States`

FREQUENTLY ASKED QUESTIONS

Frequently Asked Questions: Adult

Common questions regarding the Adult dataset, machine learning task formulations, and in-browser tabular inference.

What is the Adult dataset used for?

The Adult dataset is an established tabular machine learning benchmark extracted from the 1994 US Census, primarily used to evaluate classification algorithms on predicting whether an individual earns more than $50,000 annually.

Can I run zero-server predictions on this dataset in Google Sheets?

Yes, Carla HQ enables zero-server in-context tabular prediction directly in spreadsheets using TabICL foundation models.

What machine learning models perform best on the Adult dataset?

Gradient boosted decision trees such as XGBoost and LightGBM alongside modern tabular foundation models like TabICL routinely achieve top-tier performance (~87% accuracy / 0.92+ ROC-AUC) on this benchmark.

LIVE EVALUATION IN GOOGLE SHEETS

Test TabICLv2 on Adult Yourself

Open the pre-loaded Google Sheet and let Carla configure the target and task for local, zero-cloud tabular machine learning.

Step 1

Launch Carla

Click once to open the spreadsheet and Carla side panel together.

Step 2

Review the Setup

Carla selects the dataset target and task from this page automatically.

Step 3

Evaluate & Predict

Run predictions and compute metrics with zero server uploads.

✨ Try Adult in Carla Install Chrome Extension ↗

Provenance & Attribution

OpenML Dataset 1590

License: PublicData Source: OpenMLOpenML Page: https://www.openml.org/d/1590

Raw Data Downloads

Download Parquet OpenML Task ↗Google Sheet ↗

← Previous DatasetAbaloneAgriculture, Forestry & Fishing All Datasets Next Dataset →Airfoil Self NoiseUCI