Non-Destructive Methods Based on Machine Learning for the Prediction of Sweet Potato Leaf Area: A Comparative Approach

Leaf area is an essential parameter for studies of plant growth and physiology and is considered one of the main parameters for agricultural production. Leaf area determination methods are fundamental to understanding and predicting crop productivity. They can be classified as destructive or non-destructive. The study evaluated the performance of five methods for predicting the leaf area of sweet potato cultivars, including simple linear regression, artificial neural networks, support vector regression, adaptive neuro-fuzzy inference system (ANFIS), and random forest (RF). The coefficient of determination (R2), relative root mean squared error (RRMSE), mean absolute error (MAE), mean absolute percentage error (MAPE), and mean bias error (MBE) were used as criteria for choosing the best methods. The standardized Ranking Performance Index (sRPI) was used to classify the proposed methods. The ANFIS method performed better than the other methods analyzed (R $^{2} =0.8315$ ; RRMSE =0.0593; MAE =6.0789; MAPE =14.5741; MBE =0.000003; sRPI =1.00). Thus, the results indicated that the ANFIS method can be used as a non-destructive alternative for predicting the leaf area of sweet potato cultivars.