Which anomaly detection method is best for my data?

It depends on the distribution; Z-score is best for normal distributions, IQR is better for skewed data, and Isolation Forest excels at high-dimensional, multi-column datasets.

Does this skill support time-series data?

Yes, it includes specialized methods like STL (Seasonal-Trend decomposition using LOESS) and rolling window analysis to handle trends and seasonality.

How does the ensemble detection method work?

The ensemble method runs multiple detection algorithms simultaneously and only flags a record as an anomaly if a user-defined number of methods agree, significantly reducing false positives.

What Python libraries are required for this skill?

The skill relies on pandas, numpy, scikit-learn (for Isolation Forest and LOF), and statsmodels (for STL decomposition).

Data Anomaly Detector

Name: Data Anomaly Detector
Author: majesticlabs-dev

bymajesticlabs-dev

•

Ciencia de Datos y ML

Detects and analyzes outliers in diverse datasets using a combination of statistical methods and machine learning algorithms.

Acerca de

The Data Anomaly Detector skill provides a suite of production-ready functions for identifying outliers across various data structures, from simple numeric columns to complex time-series and high-dimensional datasets. By integrating industry-standard algorithms like Isolation Forest, Z-score, and STL decomposition, it enables data engineers and analysts to automate data quality checks, identify fraudulent activity, and monitor system health. The skill includes a robust ensemble approach that minimizes false positives by requiring consensus across multiple detection methods, making it ideal for critical production monitoring and data cleaning workflows.

Características Principales

Advanced machine learning models for high-dimensional data like Isolation Forest and Local Outlier Factor (LOF).
Customizable ensemble detection that combines multiple methods for high-precision results.
Statistical detection methods including Z-score, Modified Z-score, and Interquartile Range (IQR).
Specialized time-series analysis tools using rolling windows and STL decomposition for seasonal data.
Seamless integration with the Python data stack, including Pandas, Scikit-learn, and Statsmodels.
19 GitHub stars

Casos de Uso

Automating data quality audits to find and flag corrupt or erroneous entries in large datasets.
Monitoring financial or transactional data to identify potential fraud or security breaches.
Analyzing IoT and sensor telemetry to detect equipment malfunctions or environmental shifts.

Acerca de

Características Principales

Advanced machine learning models for high-dimensional data like Isolation Forest and Local Outlier Factor (LOF).
Customizable ensemble detection that combines multiple methods for high-precision results.
Statistical detection methods including Z-score, Modified Z-score, and Interquartile Range (IQR).
Specialized time-series analysis tools using rolling windows and STL decomposition for seasonal data.
Seamless integration with the Python data stack, including Pandas, Scikit-learn, and Statsmodels.
19 GitHub stars

Casos de Uso

Automating data quality audits to find and flag corrupt or erroneous entries in large datasets.
Monitoring financial or transactional data to identify potential fraud or security breaches.
Analyzing IoT and sensor telemetry to detect equipment malfunctions or environmental shifts.