Modeling and Predicting Sustainable Clothing Purchase Behavior of Bulgarian Gen Z Consumers

Sustainable clothing purchase behavior (SCPB) is becoming increasingly important, underscoring the need to better understand the factors that influence it. This study addresses this issue among Bulgarian Gen Z consumers by extending the theory of planned behavior with social media influence (SMI) and environmental concern (EC) and by combining explanatory and predictive analytical approaches. Data were collected from 500 respondents aged 16–28 and analyzed using partial least squares structural equation modeling (PLS-SEM) and machine learning. The measurement and structural models were evaluated, including direct, indirect, and total effects, while five regression algorithms were compared using nested cross-validation. SHapley Additive exPlanations (SHAP) analysis was subsequently applied to interpret the best-performing random forest models. The results show that perceived behavioral control (PBC) is the strongest direct predictor of purchase intention and the dominant predictor in the machine learning model for intention. Purchase intention, in turn, strongly predicts self-reported SCPB. For behavior prediction, predictive importance is more evenly distributed across PBC, purchase intention, SMI, and EC. The study contributes by integrating PLS-SEM with explainable machine learning, thereby distinguishing theoretically specified structural relationships from out-of-sample predictive importance and providing a more comprehensive understanding of sustainable consumer behavior among Generation Z.