Interaction Behavior and Entrepreneurial-Potential Assessment: Behavioral Feature Dataset and Analysis Pipeline

This dataset and analysis pipeline accompany a study examining interaction behavior in an AI-based entrepreneurial-potential self-assessment platform used in university entrepreneurship education. The platform presents 18 Likert-scale items across six trait dimensions (risk-taking, creativity/innovation, proactiveness, perseverance, autonomy, and communication/networking) and logs the full interaction sequence behind each response. The raw platform export contained 4,465 interaction events, reconstructed into 144 assessment sessions and cleaned to N = 122 valid sessions. This deposit includes: (1) the de-identified, session-level behavioral feature dataset (response timing, navigation, revision patterns, and response-style indices engineered from the interaction log); (2) the full statistical results package (descriptive statistics, reliability, exploratory and confirmatory factor analysis, correlation, regression, group-comparison, and clustering outputs); (3) the session-by-session data-quality/cleaning log documenting all attrition decisions; and (4) an executable analysis notebook implementing every feature-engineering step and statistical test reported in the associated paper, to support full reproducibility. Raw interaction-log data and any personally identifiable information are not included, to protect participant privacy, consistent with the platform's terms of use. Only aggregated, de-identified, session-level behavioral variables are shared.