Automated Financial Statement Anomaly Detection System

Develop an advanced anomaly detection system using pandas and scikit-learn that analyzes corporate financial statements, identifying potential accounting irregularities with machine learning techniques. Create a multi-stage pipeline that ingests XBRL financial data, performs dimensional reduction, applies unsupervised clustering algorithms, and generates detailed forensic reports. Implement adaptive thresholds that learn from historical financial data and can detect both subtle and significant deviations.