Prediction of Faba bean (Vicia faba L.) Yield under Artificial Shade Constraints in Moroccan Agro-fields

Background: Statistical regression models represent alternatives to process-based dynamic models for predicting the response of crop yields to climatic variations. This study chose stepwise and Lasso regression methods to predict Faba bean (Vicia faba L.) yield under different shading conditions in Moroccan agroecosystems. Methods: Crop-related data were collected in six faba bean varieties that were widely grown in full sun (S1) and two shade treatments S2 (50%) and S3 (90%) two growing seasons. The same data set was used for both regression model development and validation. Result: The results show that models based on the Lasso method were more accurate and precise (R-squared = 0.95; Root Mean Square Error (RMSE), = 102.6; Mean Absolute Percentage Error (MAPE). = 3.28%) than the stepwise approach (R²overall = 0.86; RMSE overall = 214.8; MAPE overall = 6.88%) in terms of both shading levels and growing seasons. Positive correlations were spotted between yield and NP (number of pods per plant), NS (number of seeds per plant) and WSS (100-seed weight per plant) parameters in the best shading levels (S2 and S3). These parameters significantly contributed to the Faba bean yield prediction. These findings advance the field by highlighting the potential of integrating modern regression techniques into breeding programs, offering a replicable framework for other crops and conditions.