time_series_anomaly_detection

What problem does it solve?

Detects anomalies in time-series data by comparing actual observations to Prophet-based forecasts, enabling early identification of unusual surges or slumps.

Core Features & Use Cases

• Prophet-based category-level forecasting across multiple groups
• Per-category anomaly indexing and structured summaries
• Use cases include monitoring sales, sensor data, and digital metrics to trigger interventions

Quick Start

Provide your time-series DataFrame with date, category, and value columns, then run detect_anomalies with a cutoff_date and prediction_end to receive anomaly summaries.