Intelligent Data-Driven Crop Recommendation Systems for Farmers: A Systematic Review and Classification

Abstract

Agriculture plays an important role in the Philippines, but Filipino farmers' reliance on traditional practices due to limited access to scientific technology, services, and guidance leaves them vulnerable to poor crop choices, income loss, climate risks, and low yields. This research study proposes developing a reliable, accessible, and climate-resilient crop recommendation approach for farmers. The objective of the study is to identify trends, technological methods, and research gaps. A systematic review with meta-analysis and secondary data analysis of carefully selected published studies is used as the population sampling, rather than actual human beings. To classify crop recommendation systems and identify trends, limitations, and research gaps, data were extracted using a standardized extraction form, categorized, and analyzed using descriptive statistics, mode identification, cross-tabulation, and qualitative thematic analysis. Out of the twenty (20) published studies selected, soil and weather with soil nutrients, pH, moisture, and important field factors constantly influencing the right choices are identified as the crucial input parameters in this study. Common AI-based systems primarily use machine learning and deep learning, deployed via cloud-based architectures, and are organized into a taxonomy based on method, data inputs, and deployment. Reliance on internet connectivity, high hardware costs, and difficult explainability are the challenges found in this study. Furthermore, unexplored areas are conducive to system deployment and also include inadequate farmer-centered design, limited generalizability, limited field validation, and insufficient local datasets. Moreover, researchers recommended using low-cost modular sensors, hybrid cloud–edge or TinyML deployment, and locally representative datasets with field validation.