Featured workflow map¶
eyeprocesspy carries a large article library because the package spans several research layers. This page is a practical map: choose the scientific question first, then enter the relevant workflow family.
Data ingestion and canonicalization¶
Use these workflows when the primary problem is getting heterogeneous exports into a defensible analysis state.
- Vendor/generic import and canonical interfaces
- Gazepoint real-export and workflow validation
- Coordinate-space registration and conversion
- Timebase/event alignment and file pairing
- Partitioned/interoperable storage
- Provenance and schema validation
Start with: API lifecycle and canonical interfaces.
Gaze, AOIs and process features¶
Use these workflows for fixation/saccade summaries, dwell, scanpaths, transitions, recurrence and AOI uncertainty.
- Preprocessing and feature extraction
- Scanpath sequence and transition models
- Probabilistic AOIs under calibration uncertainty
- Compositional AOI representations
- Process dynamics and contextual structure
- Presentation sensitivity and negative controls
Visual entry point: Visual gallery.
Pupillometry and multimodal process measurement¶
Use these workflows for baseline correction, functional pupil representations, missingness, synchronized process streams and staged multimodal analysis.
- Pupil preprocessing and missingness
- Functional pupil modeling
- Multimodal staged workflows
- Quality, clock and sampling audits
- Process reliability and calibration uncertainty
Start with: Calibration uncertainty and data quality.
Psychometrics and IRT¶
Use these workflows when behavioral/process features become inputs to an explicit measurement model.
- IRT foundations and diagnostics
- Process-informed IRT
- Dynamic and advanced process IRT
- Bayesian/3PL diagnostics
- DIF, score uncertainty and adaptive traces
- Cognitive-diagnosis and multidimensional extensions
Start with: Bayesian and 3PL process diagnostics.
Validation and robustness¶
Use these workflows before trusting a new estimator, feature family, or measurement claim.
- Simulation/recovery
- SBC-style evidence
- Stress testing
- Negative controls
- Grouped/leakage-aware validation
- Validation atlases and evidence programs
- Robustness and freeze artifacts
Start with: Advanced model validation.
Reproducibility and software evidence¶
Use these workflows when preparing a manuscript, package release, methods appendix, or computational audit.
- Deterministic benchmark reproduction
- Provenance manifests
- Software-paper evidence
- Release validation
- Frozen-R parity evidence
- Documentation and API audits
Start with: Benchmark reproduction.
Measurement intelligence¶
Use these workflows for higher-level measurement-system questions rather than raw gaze description.
- Process-measure registry and guardrails
- Reliability profiles
- Cross-device linking
- Calibration uncertainty
- Fairness/DIF drift
- Norms and item-bank optimization
- Measurement-intelligence utilities
These methods are intentionally accompanied by interpretive guardrails: operational usefulness, predictive performance and reliability are not substitutes for construct validation.