Predictive Anomaly Detection System for Time Series

Build an advanced anomaly detection framework for time-series data using statistical and machine learning techniques in JavaScript. Implement multiple detection algorithms including Z-score, Interquartile Range (IQR), and LSTM-based neural network approaches. Create a flexible system that can adapt to different data distributions, handle seasonality, and provide configurable sensitivity levels. Include comprehensive visualization of detected anomalies, with drill-down capabilities and exportable reports showing statistical significance and potential root causes.