Extract Data Pipeline processors

[Hugo-W]

Extract Data Pipeline processors

Description

Data transformation logic is currently scattered across models.py and utils.py. Need to unify preprocessing into a composable pipeline system.

Tasks

• Create DataProcessor base class
• Implement concrete processor classes:
  • LagProcessor (consolidate lag_matrix, lag_span, lag_sparse logic)
  • InterceptProcessor (add/remove intercept column)
  • ValidSamplesProcessor (track and apply valid sample masks)
  • NormalizationProcessor (zscore, whitening, etc.)
• Create Pipeline class to compose multiple processors in sequence
• Update TRFEstimator and other models to use the pipeline system
• Add validation and logging at each processing step

Acceptance Criteria

• Clear data transformation flow
• Reusable processors for multiple models
• Easy to add validation/logging
• Better testability of individual transformations
• Improved reproducibility

Priority

Medium

Dependencies

• Issue Move root solver.py into pyeeg/solvers.py, consolidate, and test #8 (Create models/ subpackage) should be completed first

Epic

Refactoring for Modularity

Related

Addresses architecture improvements from analysis