Machine learning for predicting sense of place among hybrid workers: A SHAP-based analysis of key factors

The rise of hybrid work has fundamentally altered how employees experience office spaces, disrupting established patterns of place-making. To identify and quantify the relative importance of key factors shaping hybrid workers' sense of place in office environments, this study employs machine learning with SHAP analysis. Based on survey data from 242 German hybrid workers, four machine learning models were compared, with the Lasso model achieving the best performance (accuracy = 0.80, AUC = 0.86). SHAP analysis identifies office environment satisfaction (SHAP value = 0.288) as the dominant factor, more than three times as influential as conscientiousness (0.087), the second-ranked factor. Additional key influential factors encompass workspace configuration (fixed desk availability, office sharing type), personality traits (neuroticism), office work percentage, and boundary management (life-to-work segmentation/integration). The findings suggest that organizations should prioritize optimizing physical office environments to enhance hybrid workers' sense of place.