Machine Learning Models to Predict Soil Moisture for Irrigation Schedule

Abstract

The agriculture industry must alter its operations in the context of climate change. Farmers can plan their irrigation operations more effectively and efficiently with the exact measurement and forecast of moisture content in their fields. Sensor-based irrigation and machine learning algorithms have the potential to facilitate farmers with significantly effective water management solutions. However, today’s machine learning methods based on sensor data necessitate a huge quantity of data for effective training, which poses a number of challenges including affordability, battery life, internet availability, evaporation issues, and other factors. The purpose of this report is to find an efficient machine learning model by doing metric evaluation and comparing the R-squared value, that can predict soil humidity within a certain crop field scenario for the next couple of days using historical data.