Upload your dataset, train a model, and get comprehensive fairness metrics — all in seconds. Ensure your AI treats every group equitably.
{
"fairness_score": 87.5,
"bias_level": "Fair",
"statistical_parity": -0.04,
"disparate_impact": 0.95,
"selection_rate": {
"privileged": 0.72,
"unprivileged": 0.68
}
}A 6-step automated pipeline from raw data to actionable fairness insights
Upload a CSV with features, a target column, and a sensitive attribute like gender or race.
Select the target column, sensitive attribute, and define the privileged group for analysis.
A Logistic Regression classifier is automatically trained on your dataset with 80/20 split.
Predictions are split by demographic group to compare outcomes across privileged and unprivileged groups.
Calculate Selection Rate, Statistical Parity Difference, and Disparate Impact Ratio per group.
A composite 0–100 fairness score combining SPD and DIR into a single actionable metric.
Upload your dataset and configure the analysis parameters
or click to browse files
No significant directional bias detected
Difference in selection rates between unprivileged and privileged groups. Closer to 0 = fairer.
Ratio of selection rates. The 80% rule requires this to be ≥ 0.80 to avoid adverse impact.
Logistic Regression coefficient magnitudes — higher = more influence on predictions
Actionable steps to improve fairness based on the detected bias patterns