FARMERS' WELL-BEING LEVEL PREDICTION: A HYBRID PLS-SEM-MACHINE LEARNING APPROACH

Abstract

Well-being is a multifaceted concept that addresses the overall quality of an individual's life. The growing literature on well-being research tends to focus on establishing relationships between various independent variables and well-being outcomes rather than solely on predictive modeling. However, more evidence is still needed to accurately predict farmers' well-being (FWB) from the social capital perspective. The current study assesses how farmers' perceptions of social capital impact well-being in rural China. A survey involving 443 farmers was conducted to scrutinize this model. The study's conceptual framework is rooted in Social Capital Theory, which types bonding, bridging, and linking social capital. The study was unique because it used hybrid PLS-SEM and Machine Learning algorithm analyses. The results indicate that linking social capital significantly influences farmers' well-being. These findings are pertinent for stakeholders in the management science sector, aiding their understanding of the importance of factors for strategic planning. Methodologically, the study contributes to the management science literature by employing a rare multi-analytical approach, including Machine Learning algorithms, to investigate FWB. The model shows that the Support Vector Regression (SVR) model with a low Mean Squared Error (0.539) has vital prediction accuracy compared to the Random Forest.

Keywords: China, farmers' well-being, social capital, PLS-SEM, Machine learning