SmartAnom is an open-source, modular low-code/no-code software framework developed for comparing and evaluating anomaly detection models. It enables users to perform data loading, synthetic data generation, model selection, hyperparameter optimization, and result visualization without requiring any programming knowledge. The system supports Isolation Forest-based methods, as well as deep learning and classical models. The Sigmoid-Based Anomaly Scoring (SBAS), located at the core of SmartAnom, is designed as an alternative to the classical scoring approach used in Isolation Forest-based algorithms. Integrated SHAP analysis enhances interpretability. Overall, SmartAnom simplifies experimentation and bridges research with real-world applications.
Serpil Ustebay (Tue,) studied this question.