Joshua Taillon - HyperSpy – Your Multidimensional Data Analysis Toolbox | SciPy 2024

HyperSpy is an open-source Python library designed for interactive analysis of multidimensional data arrays, with particular strength in materials characterization and electron microscopy

The library provides robust tools for:

• Visualization of multi-dimensional datasets
• Signal processing and analysis
• Curve and model fitting
• Basic machine learning capabilities
• Lazy processing with Dask integration
• Metadata preservation and tracking

Core concepts include:

• Division between navigation and signal axes
• NumPy-like syntax for data manipulation
• Extensible architecture through subclassing
• Integration with the scientific Python ecosystem

Notable features:

• Interactive visualization in Jupyter environments
• Built-in support for PCA, NMF, and clustering
• Customizable model fitting capabilities
• Comprehensive metadata handling
• Support for large datasets through Dask

Community aspects:

• Over 60 contributors across 14 years
• Regular tutorials and training sessions
• Focus on documentation and code quality
• Community-driven development model
• Domain-agnostic core with specialized extensions

Recent developments:

• 2.0 release focused on modularization
• Separation of domain-specific code into extensions
• Enhanced support for different data formats
• Improved scalability for large datasets
• Growing adoption beyond electron microscopy