Machine Learning–Based Discovery of Hidden Behavioral Profiles Underlying High-Performing Innovation Teams
International Journal of Innovation Management and Organizational Behavior · 2026 · 6 (3) · 1-9
DOI: 10.61838/kman.ijimob.5258 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Innovation Management and Organizational Behavior, 2026. Verfasst von Gabriela Torres und 2 weiteren Autor:innen.
Abstract (Originalsprache)
Objective
The objective of this study was to identify latent behavioral profiles underlying innovation teams and examine their predictive relationship with innovation performance using advanced machine learning techniques. Methods and Materials: This quantitative cross-sectional study was conducted among 574 employees from 84 formally designated innovation teams across technology, engineering, biotechnology, finance, and advanced manufacturing organizations in Singapore. Data were collected using validated multi-item instruments measuring collaborative orientation, psychological safety, knowledge sharing, intrinsic motivation, proactive problem solving, learning orientation, adaptive flexibility, creative self-efficacy, and collective efficacy. Team innovation performance was assessed through both leader-rated evaluations and objective performance indicators. A multi-phase machine learning pipeline was implemented, including data normalization, dimensionality reduction, unsupervised clustering using Gaussian mixture models, and supervised prediction using Random Forest and XGBoost algorithms. Model robustness was evaluated through five-fold cross-validation, out-of-sample testing, and feature importance analysis using SHAP values.
Findings
Unsupervised learning revealed four statistically distinct behavioral profiles that significantly differed in innovation performance. XGBoost achieved the highest predictive accuracy (93%), with strong discriminative power (AUC = 0.96). The Synergistic Innovators profile demonstrated significantly higher innovation outcomes than all other profiles (p < .001). Feature importance analysis identified intrinsic motivation, collaborative orientation, psychological safety, proactive problem solving, and knowledge sharing as the strongest predictors of innovation performance.
Conclusion
The findings demonstrate that innovation performance emerges from coherent behavioral configurations rather than isolated behavioral factors, and that machine learning provides a powerful framework for uncovering and predicting these hidden structures within organizational teams.
Zitieren
APA
Torres, G., Benali, F. Z., & Al-Mansoori, A. (2026). Machine Learning–Based Discovery of Hidden Behavioral Profiles Underlying High-Performing Innovation Teams. International Journal of Innovation Management and Organizational Behavior, 6(3), 1-9. https://doi.org/10.61838/kman.ijimob.5258
Vancouver
Torres G, Benali FZ, Al-Mansoori A. Machine Learning–Based Discovery of Hidden Behavioral Profiles Underlying High-Performing Innovation Teams. International Journal of Innovation Management and Organizational Behavior. 2026;6(3):1-9. doi:10.61838/kman.ijimob.5258
BibTeX
@article{Torres2026,
author = {Torres, Gabriela and Benali, Fatima Zahra and Al-Mansoori, Amina},
title = {Machine Learning–Based Discovery of Hidden Behavioral Profiles Underlying High-Performing Innovation Teams},
journal = {International Journal of Innovation Management and Organizational Behavior},
year = {2026},
volume = {6},
number = {3},
pages = {1--9},
doi = {10.61838/kman.ijimob.5258}
}
EUVIMED