Development of a Machine Learning Based Mobile and Web Crop Recommendation System for Precision Farming in Wardha

Abstract

Agricultural productivity is crucial for economic growth of a nation, especially for countries like India which is highly dependent on agriculture. This study addresses the need for cutting-edge technology to enhance crop productivity by proposing a Mobile/Web-based precision farming system. Leveraging Machine Learning (ML) algorithms, the proposed system recommends optimal crops based on several factors such as environmental parameters (Nitrogen (N), Phosphorus (P), Potassium (K)), pH value, Temperature, Rainfall, Humidity and Soil Nutrients., The current study applied five ML algorithms Naïve Bayes, Random Forest, on a dataset comprising of 2200 records and 7 variables to create a predictive ML model. Among the five algorithms, the Naïve Bayes exhibits superior performance, achieving a precision rate of 99.50%. Despite challenges like data scarcity and the need for field validation, the research aims to provide Wardha farmers with a free and open-source precision farming platform. By facilitating informed crop management decisions, the platform seeks to augment agricultural productivity and foster sustainable economic growth in the region. The robustness of the proposed system is further validated by the identified algorithmic accuracy, with Random Forest emerging as the most efficient option for crop recommendation within the specified environmental context. This study under-scores the potential of precision farming to revolutionize agriculture and contribute to long-term economic development in emerging regions. The validation process involves a train-test split, cross-validation, and field validation to ensure the reliability and accuracy of the machine learning models. Additionally, the impact assessment highlights potential benefits such as improved crop yields, resource optimization, sustainable economic growth, and long-term environmental benefits for the Wardha region.