Paper Contents
Abstract
The integration of artificial intelligence (AI) and data-driven methodologies into the agricultural sector has led to a surge in smart farming practices aimed at enhancing crop yield predictions. This paper presents a novel approach that combines automated code review systems with predictive modelling techniques for crop yield estimation using remote sensing and meteorological data. Traditionally, the development of such predictive models involves significant manual intervention, particularly in the implementation, validation, and refinement of code. By introducing automated code review systems into the model development workflow, we streamline the process of ensuring code quality, correctness, and efficiency, thus accelerating the deployment of robust predictive systems.Our study focuses on leveraging satellite-based remote sensing datasuch as Normalized temperature, and soil moisturealongside meteorological data, including rainfall, temperature, humidity, and solar radiation. (LSTM) Networks for multi-temporal crop yield prediction across varying agro-climatic zones.We incorporate an automated code review system within the model development pipeline to enforce best practices in coding, detect potential bugs, and validate algorithmic logic. This system uses static analysis, linting tools, and machine learning-based code quality checkers to review code submissions in real time. The automated review not only ensures maintainability and reproducibility of the models but also reduces development cycle time and human error.A case study was conducted on maize and wheat crops in select regions of Sub-Saharan. Results show that integrating automated code reviews significantly improved the consistency and reliability of model outputs. The best-performing model achieved an R of 0.87 for maize and 0.82 for wheat, demonstrating high accuracy in yield prediction. Furthermore, the automated review process identified several instances of suboptimal code and logic errors that would have otherwise gone unnoticed until later validation stages.Immense potential in addressing food security challenges. This research not only underscores the importance of accurate data sources and advanced modeling techniques in yield prediction but also advocates for the integration of software engineering best practicesespecially automated code reviewin agritech applications. Future work will explore the incorporation of real-time feedback loops, continuous integration pipelines, and adaptive learning models that evolve with incoming data streams.
Copyright
Copyright © 2025 Amrendra Kumar. This is an open access article distributed under the Creative Commons Attribution License.
