XGBoost-Based Prediction of Innovative Work Behavior Using Organizational Climate and Leadership Variables

Abstract

Objective: The present study aimed to develop and evaluate an XGBoost-based machine learning model for predicting innovative work behavior using organizational climate and leadership variables among employees working in Indonesian organizations.

Methods and Materials: This study employed a cross-sectional predictive research design involving 1,248 employees from multiple Indonesian organizations representing manufacturing, information technology, telecommunications, financial services, and public administration sectors. Data were collected using standardized measures of innovative work behavior, organizational climate, and leadership. Demographic and occupational variables were also included as supplementary predictors. Data preprocessing procedures included missing-value treatment, standardization, and quality screening. The dataset was divided into training (80%) and testing (20%) subsets. Extreme Gradient Boosting (XGBoost) was utilized to develop the predictive model, while hyperparameter optimization was performed using five-fold cross-validation and grid-search procedures. Model performance was evaluated using the coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). SHapley Additive exPlanations (SHAP) analysis was applied to identify and interpret the relative importance of predictor variables.

Findings: Inferential analyses revealed significant positive correlations between innovative work behavior and organizational climate (r = .71, p < .01), transformational leadership (r = .68, p < .01), transactional leadership (r = .41, p < .01), and organizational tenure (r = .19, p < .01). The XGBoost model demonstrated strong predictive performance, explaining 88.7% of the variance in innovative work behavior in the training dataset and 84.2% in the testing dataset (R² = .842). The model achieved low prediction errors, with testing RMSE = 0.287, MAE = 0.223, and MAPE = 7.46%, indicating excellent predictive accuracy and generalizability. SHAP analysis identified transformational leadership as the most influential predictor, followed by innovation and flexibility climate, supervisory support, autonomy, communication quality, organizational integration, contingent reward leadership, educational level, organizational tenure, and organizational size. The results demonstrated that leadership and organizational climate variables accounted for the majority of predictive power within the model.

Conclusion: The findings demonstrate that innovative work behavior can be predicted with high accuracy using organizational climate and leadership variables. Transformational leadership and innovation-supportive organizational climates emerged as the most influential determinants of employee innovation. The study highlights the value of explainable machine learning techniques for organizational research and provides evidence that leadership development and innovation-oriented workplace environments represent critical levers for enhancing innovative work behavior.