Correlation analysis between pathfinding factors and spatial attributes
Description
Table 1: Wayfinding Performance at Key Nodes Core content: Average errors and error rates at 8 key nodes of Xizhimen Station. Variables: Node ID (unique identifier), Average Number of Errors (mean errors per node across 30 participants), Wayfinding Error Rate (calculated as [Average Errors/(1+Average Errors)]×100). Key feature: Errors concentrated at Node 3 (25.00%), 4 (37.50%), 6 (40.00%); others show 0% error. Table 2: Spatial Attributes and Wayfinding Performance Core content: Integrates independent variables (Spatial Connectivity/Visual Area) and dependent variables (Avg. Decision Time/Error Rate) for 8 nodes. Variables: Spatial Connectivity (number of directly connected units), Visual Area (m², isovist polygon area), Avg. Decision Time (s), Error Rate (%). Key feature: High connectivity/visual area nodes (4,6) correlate with higher error rates and longer decision times. Table 3: Correlation (Error Rate vs. Spatial Attributes) Core content: Pearson correlation results (n=8, two-tailed α=0.05). Results: SC-ER (r=0.411, p=0.311), VA-ER (r=0.414, p=0.308) — weak positive correlations, not statistically significant due to small node sample. Table 4: Correlation (Decision Time vs. Spatial Attributes) Core content: Pearson correlation results (n=8, two-tailed α=0.05). Results: SC-DT (r=0.513, p=0.194), VA-DT (r=0.510, p=0.197) — moderate positive correlations, near significance. Table 5: Grouped Analysis (Spatial Connectivity) Core content: Detailed SC, DT, ER for each node. Role: Input data for SC-DT/SC-ER cubic regression equations. Table 6: Predicted vs. Measured ER (Spatial Connectivity) Core content: Compares measured ER, regression-predicted ER, and residuals. Key validation: High accuracy (e.g., Node 6: predicted 40.12%, measured 40.00%), confirming model reliability. Table 7: Grouped Analysis (Visual Area) Core content: Detailed VA, DT, ER for each node. Role: Input data for VA-DT/VA-ER cubic regression equations. Table 8: Predicted vs. Measured ER (Visual Area) Core content: Compares measured ER, regression-predicted ER, and residuals. Key validation: Low prediction deviation, supporting nonlinear relationship validity. Table 9: Visual Area Safety Threshold Core content: Classifies nodes as "Safe" (ER≤20%)/ "Risk" (ER>20%) based on 538–955 m² threshold. Key criterion: Nodes within threshold are mostly safe (except Node 3), verifying threshold effectiveness.
Files
Steps to reproduce
2. Steps of Reproduction 1.Participant Recruitment: Calculate sample size via G*Power 3.1 (f=0.25, α=0.05, power=0.95) — recruit 30 healthy adults (15M/15F, 22.4±2.1 years), screen for normal vision and no Xizhimen Station experience, obtain informed consent. 2.Virtual Environment & Node Selection: Build 1:1 sign-free VR model of Xizhimen Station. Select 8 nodes per functional criticality, spatial attribute representativeness (SC:346–1181, VA:341.9–1180.2 m²), and transfer line coverage. Quantify SC/VA via DepthmapX. 3.Experiment & Data Collection: 10-min preparation (VR headset + heart rate monitor, 3-min baseline); 5-min free exploration; 15-min standardized wayfinding (5 routes ×3 repeats). Collect behavior data (90 Hz) and eye-tracking data (120 Hz). 4.Data Preprocessing: Code wayfinding results (0=correct,1=suboptimal,2=incorrect); calculate Average Errors and Error Rate; average 30 participants’ data per node (n=8); conduct Grubbs test (z<1.85, p>0.05) to exclude outliers. 5.Statistical Analysis: Use SPSS 27 for Pearson correlation (Tables 3–4); fit cubic regression equations via OriginPro 2025b and SPSS 27 (correct rounding errors) (Tables 5–8); perform ROC analysis via R v4.1.0 (proc package) to determine visual area threshold (Table 9); visualize with ggplot2. 6.Quality Control: Cross-verify data synchronization; validate model prediction deviation (<2% between software).
Institutions
- Beijing Jiaotong University