What is TanaT?#
TanaT (Temporal ANalysis of Trajectories) is an extensible Python library for temporal sequence analysis with a primary focus on patient care pathways.
The name also refers to a variety of wine grape that originates from south of France, taking continuity with the TraMineR library which widely inspired this work (Traminer is also a variety of wine grape).
Key Features#
- Expressive Data Representation
TanaT provides a very expressive and flexible representation of event-based temporal data, distinguishing between entities, sequences, and trajectories.
- Multiple Sequence Types
Support for event sequences (point-in-time), interval sequences (duration-based), and state sequences (continuous states).
- Advanced Analytics
Implements different metrics and clustering algorithms specifically designed for temporal sequence data.
- Extensible Architecture
Built with extensibility in mind, making it easy to add new metrics, clustering methods, and analysis techniques.
What Makes TanaT Different?#
Unlike traditional time series libraries, TanaT is designed for irregularly sampled, symbolic event-based temporal data. We classify timed sequences data as non-euclidean in the sense that there is no natural euclidean space to represent collections of timed sequences.
For instance:
Timed sequences in a pool don’t necessarily have the same length (same number of events)
Events themselves have no natural euclidean representation (when using symbolic event types)
Traditional data analysis methods don’t apply directly to these non-euclidean data
Core framework functionalities#
The TanaT framework provides a complete workflow for temporal sequence analysis, from data ingestion to advanced analytics and visualization.
- Simulation
Synthetic data generation for statistical power analysis and comprehensive benchmarking of algorithms
- Visualization
Rich visualization tools for exploring and interpreting temporal sequences and analysis results
- Data Wrangling
Flexible data manipulation, filtering, and transformation capabilities
- Survival Analysis
Integration with survival analysis techniques for time-to-event data
- Metrics & Clustering
Distance metrics and clustering algorithms specifically designed for temporal sequences
- Workflow Orchestration
Pipeline management for reproducible and automated analysis workflows