This study explores the feasibility of predicting cucumber growth in smart greenhouses using routinely collected environmental data. Weekly plant growth metrics—including height, stem diameter, and fruit count—were analyzed in conjunction with high-frequency sensor data for CO₂ concentration, internal air temperature, and root-zone temperature. The data were sourced from a commercial smart greenhouse in South Korea. Multiple predictive models were tested, including statistical methods like Multiple Linear Regression and machine learning algorithms such as Decision Tree Regression and Support Vector Regression. Across all models, predictive accuracy was poor, with negative R2 values indicating that the environmental variables used were inadequate for modeling growth outcomes. These findings underscore the limitations of relying solely on a narrow set of environmental sensors in precision agriculture. The study advocates for the integration of additional parameters—such as light intensity, nutrient solution composition, and plant physiological indicators—to improve model reliability. Ultimately, the results highlight the importance of combining rich, multidimensional data with advanced modeling techniques to build robust AI-driven cultivation systems in smart farming environments.