Dataset Attribute Description Details for Risk-Based Authentication Research

A 5.5 KB Excel file describing attributes for a study on Risk-Based Authentication (RBA). The dataset supports a hybrid RBA framework integrating ensemble machine learning, fuzzy logic, clustering, and optimization for account takeover detection. Authored by K. Sasikumar and last updated on 2026-05-26, it is licensed under CC-BY-4.0.

Use Cases

• Training ensemble classifiers for account compromise prediction based on login behavior, device attributes, and network information.
• Setting and refining risk thresholds for authentication decisions using K-Means clustering and fuzzy logic.
• Applying Explainable AI techniques like SHAP to interpret model predictions for risk-based authentication.
• Benchmarking authentication framework performance on large-scale datasets ranging from 2M to 30M records.

Strengths

• The described framework achieved high performance metrics, including 97.77% accuracy and a 0.0303 EER.
• Experimental results show consistent improvement on large-scale datasets, with specific action counts reported for a 2M record dataset.
• The integration of Explainable AI (SHAP) is noted to enhance model transparency and interpretability.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

figshare

Freshness

Last updated 2026-05-26 17:44:49; freshness should be verified.