De-identified dataset and analysis script for: Evaluating the affective and restorative performance of building interiors with virtual reality and multimodal measures (memorial transition space feasibility study)

De-identified per-participant dataset (N=12, coded P01-P12) and a dependency-free Python analysis script accompanying a pilot feasibility study of a reproducible virtual-reality and multimodal protocol for evaluating the affective and restorative performance of building interiors. Version 2 adds raw per-participant data (per-second EEG streams and item-level PANAS/STAI-S/PRS responses) and corrects the Section 3.6 power analysis in analysis.py. Contents: merged_subj_eeg_deid_en.csv - self-report scales (PANAS, STAI-S, PRS), EEG-derived indices, session duration, demographics; English labels (30 columns) - merged_subj_eeg_deid.csv - identical, Chinese demographic labels (30 columns) - eeg_per_subject_deid.csv - per-participant EEG-derived index summaries + duration (18 columns) - P01_eeg.csv to P12_eeg.csv - per-participant per-second EEG index streams (attention/emotion/workload, ~1 Hz, relative time) - questionnaire_items_deid.csv - item-level responses (12 x 62 items: PANAS 20, STAI-S 20, PRS 22) - T01_trajectory.csv to T12_trajectory.csv - head-position trajectories (~10 Hz) for locomotion visualization (separate T-code index) - analysis.py - stdlib-only script reproducing all reported effect sizes, correlations, and power analyses - data_deidentified.pdf - data dictionary / codebook and de-identification notes Note: EEG indices are vendor-derived from a consumer-grade wearable headband; construct validity is not independently verified. Age is binned into 5-year bands for privacy. No significance testing is reported (n=4 per group); effect sizes serve as planning inputs for a future powered study.