Predictive Model for Smart Agriculture Using Machine Learning

Abstract

Agriculture holds a significant position in India's economic and employment landscape. A common challenge faced by Indian farmers is the lack of adherence to appropriate crop selection based on soil requirements, resulting in the cultivation of crops without a well-structured Crop Recommendation System. This adversely affects the productivity of crop yields. Precision agriculture emerges as a solution to this issue, characterized by the utilization of a soil database derived from the farm, expert-provided crop recommendations, and the incorporation of parameters such as soil quality from soil testing lab datasets. The proposed system takes input data on soil quality, including Nitrogen, Phosphorous, Potassium, and pH values, as well as weatherrelated information such as rainfall, temperature, and humidity. This information is utilized to predict the optimal crop for cultivation and enhance crop productivity. The research employs datasets obtained from the Kaggle website, utilizing machine learning algorithms to analyze the data. The study focuses on key parameters to determine the most suitable crops for cultivation in specific regions, aiming for heightened productivity. Among the various machine learning algorithms applied, both the Random Forest and Naive Bayes Algorithmsmdemonstrated comparable results, achieving a remarkable accuracy score of 99.09% although XG Boost exhibited the highest accuracy, with a score of 99.31%.