eyeprocesspy

Reproducible Python infrastructure for eye-tracking, pupillometry, AOIs, process data, psychometrics, and multimodal behavioral measurement.

Documentation · Getting started · Visual gallery · 88 workflow articles

eyeprocesspy is the Python companion and deep-parity port of the R package eyeprocess, with frozen eyeprocess 0.11.1 as the scientific reference. It brings vendor import, canonical data contracts, preprocessing, gaze/AOI analysis, pupil workflows, process measurement, IRT, validation, scientific plots, provenance, and reporting into one auditable package.

0.1.0 release evidence: the controlling deep-parity gate passed with 1,458 tests, 23,085 / 23,085 statements, and 9,680 / 9,680 branches covered. The frozen API and article ledgers are complete, and the cross-platform release matrix is green.

Release snapshot

The exact evidence is recorded in RELEASE_VALIDATION.md and TEST_SUMMARY.md.

Installation

After publication to PyPI:

pip install eyeprocesspy

For development or source installation:

pip install "git+https://github.com/stefanosbalaskas/eyeprocesspy.git@v0.1.0"

Windows manual installation

The hardened installer has been exercised successfully on a real Windows installation with Python 3.11.9. Package verification passed, the recommended extras installed, and python -m pip check reported No broken requirements found.

From the extracted manual-install bundle:

Set-ExecutionPolicy -Scope Process Bypass
.\install_eyeprocesspy.ps1 -WithAllRecommended

The installer does not require the Windows py launcher and can also use an explicitly supplied interpreter path. See the manual-install guide.

Why eyeprocesspy?

• One coherent data model for recordings, gaze samples, eye/pupil samples, fixations and episodes, events, intervals, AOIs, responses, features, quality, and provenance.
• Scientific parity first: 1,182 / 1,182 frozen APIs are resolved against eyeprocess 0.11.1, with governed records for unavoidable cross-language differences.
• Process data as first-class evidence: scanpaths, transitions, temporal structure, uncertainty, reliability, and psychometrics live in the same analytical surface.
• Measurement guardrails: calibration uncertainty, quality, reliability, DIF/fairness, and process metrics retain explicit interpretation boundaries.
• Reproducibility by construction: deterministic benchmarks, provenance, validation evidence, software-paper evidence, and release audits are built in.
• Broad scientific plotting surface: gaze, AOI, pupil, quality, IRT, process-measurement, validation, and model-diagnostic graphics are supported through Matplotlib-oriented workflows.

Visual tour

Gaze trace	Scanpath

Pupil time series	Probabilistic AOI membership

Process reliability	IRT information

Open the complete visual gallery →

30-second reproducible check

import eyeprocesspy as ep

study = ep.eyeprocess_benchmark_study()
audit = ep.validate_benchmark_study(study)
data = ep.import_benchmark_study(study)

print(audit["valid"])
print(data)

For a real export:

import eyeprocesspy as ep

eye = ep.read_eye_export("participant_001.csv", vendor="auto")
issues = ep.validate_eye_dataset(eye)

Capability map

Documentation

• Website: https://stefanosbalaskas.github.io/eyeprocesspy/
• Getting started
• Manual installation
• Runnable examples
• Practical cookbook
• Visual gallery
• Python-native guides
• 88-article workflow library
• API and plotting reference
• FAQ
• Parity and validation
• Release and reproducibility

Scientific boundary

eyeprocesspy provides measurement and analysis infrastructure. A metric is not automatically a validated psychological construct, diagnosis, or causal explanation. Reliability does not establish construct validity; prediction does not establish causation; probabilistic AOI membership reflects modeled coordinate uncertainty rather than probability of attention; and gaze, pupil, and biometric measures require an appropriate design, measurement model, and ethical interpretation.

Relationship to R eyeprocess

The Python package is developed against the frozen eyeprocess 0.11.1 reference. API, articles, data, plots, backends, numerical evidence, and unavoidable language-specific divergences are tracked explicitly. Python-native extensions are separated from reference parity so they do not masquerade as R-equivalent behavior.

License

See LICENSE.