CONSTRUCTION AND VALIDATION OF A PREDICTIVE MODEL FOR TOMATO ORGAN BIOMASS AT ORGAN SCALE BASED ON STACKING LEARNING

In this study, a stacked machine learning algorithm was constructed with tomato organ biomass as the research object, taking the geometric morphology data of tomato organs as the variables, utilizing eight classical machine learning algorithms as the base-model, and applying the linear regression algorithm as the stacked meta-model. This algorithm was then utilized to establish a prediction model for tomato biomass at the organ scale, and the biomass models of tomato plant leaves and fruits at the organ scale were constructed. The model has R2=0.86, MAE=0.49, and RMSE=0.81 in predicting leaves, and R2=0.94, MAE=0.33, and RMSE=0.57 in predicting fruits. The model has practical applications in predicting tomato yield and supply, providing market information, and supporting agricultural investment decisions. It also helps to optimize agricultural production and management, guide industrial development and planning, and improve the efficiency and competitiveness of the agricultural sector.