A Review of Challenges in Applying Machine Learning to Agricultural Big Data

Abstract

Agricultural Big Data consists of various technologies that can help with the challenges of the new data era. When used together with machine learning methods, agricultural data can support farmers in making decisions, managing water, soil, crops, and livestock. Applications related to crop management include predicting yields, identifying diseases, weeds, assessing crop quality, and identifying species. Livestock management deals with improving animal welfare and productivity.

This paper aims to summarize the issues related to using machine learning in Agricultural Big Data systems. A systematic literature review was conducted using the PRISMA protocol, including 30 research articles published between 2015 and 2020.

Our proposal is based on the findings and presents a framework highlighting key challenges, machine learning methods, and technologies used. The design of the Agricultural Big Data architecture is considered one of the most important challenges, as it constantly changes with new machine learning methods and data volumes.

Keywords: Smart Agriculture, Crop Yield Prediction, Machine Learning, Deep Learning, Precision Farming, IoT, Agricultural Big Data