Home/Datasets/Network-Based Anomaly Detection Dataset

Network-Based Anomaly Detection Dataset

A structured dataset for network-level anomaly detection in electricity distribution systems using feeder and distribution transformer (DT) level operational data.

About Dataset

The Network-Based Anomaly Detection Dataset is a structured compilation of monthly operational data from electricity distribution networks at feeder and distribution transformer (DT) levels for the period of 11 months. The dataset includes key parameters such as input energy, billed energy, collection efficiency, AT&C losses, consumer mix, connected load, and DT capacity and loading characteristics. Derived features have been engineered to capture energy balance, loss intensity, infrastructure utilization, and revenue leakage, enabling advanced analytics without requiring sensitive high-frequency load curves or theft records. The dataset is designed to support anomaly detection, risk scoring, and operational diagnostics across power distribution networks. It allows researchers and practitioners to develop machine learning models for identifying inefficiencies, inconsistencies, and potential anomalies in network behavior while preserving data privacy constraints.

Purpose of Dataset

This Dataset Is Intended To Support The Development Of Ai/ml Models For Anomaly Detection In Power Distribution Networks. It Can Be Used For Unsupervised Anomaly Detection, Risk Scoring Of Feeders And Transformers, Infrastructure Utilization Analysis, And Benchmarking Of Network Efficiency. It Is Particularly Suited For Environments Where Sensitive Data Such As Theft Records Or High-frequency Load Curves Are Not Available.