Package Proposal for Data Pre-Processing for Machine Learning Applied to Precision Irrigation

Abstract

The evolution of the Internet of Things (IoT) devices for precision agriculture is directly linked to the needs and interests of humanity. These advances include migration to cloud computing, data engineering, and the democratization of tools. These changes allow for better management, data quality, security, and scalability, reducing operational costs. The objective of this research was to present a proposal for a data pre-processing package for meteorological stations classified as conventional. Among the main findings of this research is the need for data pre-processing for Machine Learning applications focused on precision irrigation, controlled by IoT devices; the use of data from conventional weather stations for Machine Learning applications; the availability of applications developed in Open Source repositories, and the proposal of a data pre-processing package to help professionals from different areas. The systematic review examined the various machine-learning applications for precision irrigation. Different models and mechanisms used to apply Machine Learning in precision irrigation projects were identified. In addition, we look at the challenges faced when using Machine Learning for precision irrigation, including the lack of data, the need for efficient data pre-processing, and the need to tune the model to get the best possible result. At the end of the article, we propose a data pre-processing package for conventional meteorological stations. This package includes normalization, noise removal, and outliers to improve the reliability of the input data.