Quality Assessment and Predictive Modeling of Chickpea, Lentil and Camelina Yield: Effects of Nitrogen, Irrigation, Rainfed Conditions, and Intercropping Using Machine Learning Approaches

Given increasing water scarcity, the need to reduce chemical inputs, and the growing demand for nutritionally valuable vegetable oils, identifying sustainable cropping systems for camelina production has become increasingly important. This study aimed to evaluate the combined effects of nitrogen application, irrigation regimes (irrigated and rainfed), and different intercropping patterns of camelina with chickpea and lentil on yield components, oil quality traits, and water productivity, and to assess the predictive capability of machine learning models for these traits. A 2‐year field experiment was conducted using a factorial split‐plot arrangement based on a randomized complete block design. Measured variables included oil yield, oil content, fatty acid composition, camelina meal protein yield, and water productivity indices. To capture complex and nonlinear relationships among management factors and measured traits, principal component analysis (PCA) was first applied for data dimensionality reduction, followed by Gaussian process regression (GPR), radial basis function neural network (RBF‐NN), and random forest (RF) models for trait prediction. The results demonstrated that intercropping camelina with chickpea or lentil, particularly under balanced planting ratios, significantly improved oil quality and water productivity compared to sole cropping. Low nitrogen application (11.5 kg ha − 1) enhanced polyunsaturated fatty acids and preserved desirable oil quality attributes, whereas higher nitrogen rates generally reduced the concentration of key fatty acids. Among the tested models, RBF‐NN exhibited the highest accuracy in predicting oil yield and major quality traits, while GPR performed effectively in simulating fatty acid composition. Overall, the findings highlight that integrating optimized agronomic management with machine learning approaches provides a powerful framework for improving oil quality, resource‐use efficiency, and sustainability of camelina‐based cropping systems in semiarid and rainfed environments.