Outlier Detection

Detect anomalies and outliers in your data using statistical methods.

Z-Score Method

Identify outliers using standard deviations:

Copy
-- Detect outliers using Z-score
SELECT id, features,
       z_score_outlier_detection(features, 3.0) AS is_outlier
FROM data_table;
-- is_outlier = true if |z-score| > 3.0

Modified Z-Score

More robust to outliers than standard Z-score:

Copy
-- Modified Z-score detection
SELECT id,
       modified_zscore_outlier(features, 3.5) AS is_outlier
FROM data_table;

IQR (Interquartile Range) Method

Detect outliers using quartiles:

Copy
-- IQR-based outlier detection
SELECT id,
       iqr_outlier(features) AS is_outlier
FROM data_table;

Isolation Forest

Tree-based anomaly detection:

Copy
-- Isolation Forest
SELECT isolation_forest_outlier(
    'data_table',
    'features',
    0.1  -- contamination rate
);

Learn More

For detailed documentation on outlier detection methods, choosing thresholds, visualization, and handling outliers, visit:

Outlier Detection Documentation

Related Topics

• Drift Detection - Detect data distribution changes
• Quality Metrics - Data quality assessment