Plot gallery¶

This gallery is generated by **gpbiometricspy itself** from bundled synthetic/public demonstration data, with the 0.1.8 PPG measurement-equity tranche adding three deterministic SVG evidence figures alongside the 17 CI-generated PNGs. The figures are not screenshots or hand-drawn mockups.

Current stable documentation for 0.1.8. Documentation CI regenerates the 17 core PNG figures from the checked-out package source. The three PPG measurement-equity SVGs are generated deterministically by `scripts/generate_ppg_equity_synthetic_demo.py` and retained as separate known-truth evidence assets. The former validator marker “Current development documentation.” is retained here only as historical wording; this page now documents the public stable 0.1.8 release.

Quality EDA / SCR PPG / HRV Pupil / gaze / AOI Alignment Design / inference

Regenerate the core gallery locally with:

PYTHONPATH=src python scripts/generate_docs_gallery.py
python scripts/validate_docs_site.py

Regenerate the PPG measurement-equity figures with:

python scripts/generate_ppg_equity_synthetic_demo.py \
  --output-dir docs/assets/ppg-equity \
  --seed 20261003 \
  --participants 36 \
  --n-boot 400

20showcased deterministic documentation figures

17 + 3core CI gallery + PPG measurement-equity figures

6major workflow domains represented visually

0private participant datasets required

From figure to workflow¶

Use a plot as an evidence checkpoint, then continue into the workflow that produced it. Each route below connects a visual domain to a worked example and to the matching shareable API lens.

Quality

Inspect quality and reporting

Start from missingness, signal activity and report-ready diagnostics.

Open filtered QC API →

EDA / SCR

Reproduce conductance diagnostics

Move from recorded conductance through decomposition and candidate response events.

Open filtered EDA API →

PPG / HRV

Reproduce cardiac diagnostics

Inspect pulse peaks, interval provenance, variability diagnostics, and the separate measurement-equity route added in 0.1.8.

Open filtered cardiac API → · Read the measurement-equity article →

Eye tracking

Reproduce pupil, gaze and AOI diagnostics

Connect validity, AOI definitions and gaze-derived visual checks.

Open filtered eye-tracking API →

Alignment

Reproduce the multimodal timeline

Inspect events and streams on an explicit shared coordinate system.

Open filtered alignment API →

Design / inference

Choose the next modelling step

Use design coverage and inferential diagnostics before adding model complexity.

Open filtered statistics API →

Interpretation boundary. A generated plot documents recorded data, processing or an inferential diagnostic. It is not, by itself, evidence of emotion, stress, trust, attention, preference, diagnosis, sensor validity, causal effects or hardware-level synchronization.

Signal overview and quality¶

Standardised EDA and heart-rate signal overview — **Biometric signal overview.** Standardised EDA and heart-rate traces from the bundled kiosk demonstration.

Missingness overview for EDA heart rate IBI and pupil channels — **Missingness overview.** Channel-level availability across EDA, heart-rate, IBI and pupil measurements.

EDA signal quality diagnostic — **Signal-quality diagnostic.** A compact core-QC summary suitable for preprocessing review.

Biometric quality overview across EDA heart-rate IBI and pupil channels — **Biometric quality overview.** A broader multi-channel view for manual QC and reporting.

Continue with the quality and reporting example or the dataset validation guide.

EDA and SCR¶

EDA tonic and phasic decomposition — **EDA decomposition.** Observed signal with tonic and phasic components.

Detected skin conductance response events — **SCR events.** Deterministically detected skin-conductance responses over the EDA trace.

EDA time frequency gram — **EDA-gram.** Time-frequency diagnostic for slow electrodermal dynamics.

See the EDA / SCR example for preprocessing, event detection, QC, and reporting context.

PPG and HRV¶

PPG waveform with detected pulse peaks — **PPG peak detection.** Pulse waveform and accepted peaks from the deterministic synthetic generator.

Poincare plot of successive beat intervals — **Poincaré plot.** Beat-to-beat geometry for interval-variability inspection.

Beat-to-beat interval tachogram — **HRV/PRV-style tachogram.** Beat-to-beat interval trajectory through the pyHRV-style visual interface; scientific naming still depends on source provenance.

Use the PPG / HRV example to keep peak detection, interval provenance, and HRV-versus-PRV interpretation explicit.

PPG measurement equity — new in 0.1.8¶

The following three SVG figures are generated by the deterministic PPG measurement-equity example. They deliberately separate waveform properties, candidate-value availability, and reference error.

Synthetic PPG waveforms used for measurement-equity signal-quality diagnostics — **Raw waveform / SQI evidence.** Acquisition-level differences are inspected before downstream agreement claims.

Synthetic candidate heart-rate retention across ITA values — **Retention over ITA.** The known-truth generator makes candidate-HR availability vary across the observed objective-pigmentation range.

Synthetic reference heart-rate error across ITA values — **Reference error over ITA.** The synthetic HR-error mechanism remains ITA-neutral even while availability changes.

Follow the PPG measurement-equity synthetic example, read Auditing PPG measurement equity, or inspect the authorized STEP external-evidence record.

Pupil, gaze and AOIs¶

Pupil diameter and gaze position overview — **Pupil and gaze overview.** Standardised pupil diameter and gaze coordinates for one synthetic participant.

AOI linked biometric summary — **AOI-linked biometrics.** EDA, heart-rate and pupil summaries grouped by areas of interest.

Saccade main sequence amplitude versus peak velocity — **Saccade main sequence.** Amplitude/peak-velocity relationship used as a gaze-event diagnostic.

Continue with the pupil, gaze and AOI example for validity, event, AOI, and pupil-specific checks.

Multimodal alignment¶

Multimodal EDA heart-rate and pupil timeline — **Multimodal timeline.** Synchronous EDA, heart-rate and pupil channels with event markers on the demonstration timebase.

The multimodal example and timebase guide separate coordinate alignment from evidence about actual hardware synchronization.

Design and inference¶

Experimental design coverage across conditions and trials — **Design coverage.** Condition/trial coverage inspected before modelling so sparse or imbalanced cells remain visible.

Two-condition time course with cluster permutation result — **Cluster-permutation time course.** A deterministic synthetic two-condition effect with detected cluster information. Cluster timing is descriptive rather than precise onset evidence.

Permutation null distribution for cluster statistic — **Cluster null distribution.** The global permutation reference distribution used by the synthetic cluster example.

For model choice, use the modelling strategy guide; for process-data inference, read the relevant method notes before interpreting timing or heterogeneity.

Read the figures as diagnostics¶

The gallery is designed to make intermediate evidence visible. A diagnostic plot is not a scientific conclusion by itself:

missingness and quality plots document recorded-data properties;
decomposition and peak plots document transformations and detection behavior;
PPG measurement-equity figures keep acquisition quality, availability, and error distinct;
gaze plots document measured or derived eye-tracking behavior;
design plots document coverage;
cluster plots summarize an inferential procedure under its stated exchangeability assumptions;
multimodal plots show aligned coordinates but do not prove hardware synchronization.

Where the code lives¶

The 17 core PNGs are generated from public package functions by scripts/generate_docs_gallery.py. The three PPG measurement-equity SVGs are generated by scripts/generate_ppg_equity_synthetic_demo.py. See the domain examples, the 26 executable frozen R-companion articles, the Python-native explanation articles, and the 406-function API reference for corresponding code and workflow context.