A Machine Learning Model for Automating Irrigation System

Abstract

The India most people are dependent on agriculture-based businesses. It also contributes to our economy. Automation in agriculture will help improve crop production quality and quantity. Additionally, help to minimize the resources required. But automation in agriculture requires a significant amount of investment. Therefore, in order to minimize the automation cost of the agricultural irrigation system a Machine Learning (ML) technique is proposed. The aim of the proposed ML model is to reduce the sensor implementation cost and running cost of the complete remote sensor network. Therefore, we considered a dataset of soil moisture and temperature dataset where the predictable is an irrigation treatment type. The dataset is pre-processed first to transform data for learning, then the k-means clustering is applied for behavioral data analysis. This process groups the similar sensor reading and enhances the learning performance regarding the accuracy and learning time. Finally, two machine learning techniques have been implemented to train and predict the irrigation treatment. This technique minimizes the expenses of implementing the sensors on the ground and their execution and maintenance costs. However, the entire system’s performance depends on the accuracy of the prediction model, therefore in near future, we need to work on improving the prediction accuracy of the proposed model.