Recommendation System for Crops Integrating with Specific soil parameters by Machine Learning Techniques

Abstract

The process of investigating a set of data to extract useful information from that set is known as “data mining.” Data mining has many applications, including in the areas of commerce, health, agriculture, and more. In order to improve the quality of production while reducing the usage of pollutants agriculture uses data mining to analyse the many different environmental factors. Agriculture is the primary contributor to India's economy and accounts for most of the country's workforce. The majority of Indian farmers have the same issue: they do not choose the most suitable crop for their land based on the needs of their environment. As a result, they will see a major decline in their total level of productivity. Precision agriculture has provided a solution to the problem that the farmers were experiencing. The term “precision agriculture” refers to a contemporary farming practice that uses collected research data on soil features and kinds, as well as crop yields, and then recommends to farmers the kind of crop that would yield the most in that specific area. This leads to a decrease in crop selection errors and an increase in yield. This paper proposes a recommendation system through an ensemble model with a majority voting technique that uses a Decision Tree, Naive Bayes, SVM, Logistic Regression, Random Forest (RF), and XGBoost as learners to recommend a crop for the site-specific parameters with high accuracy and efficiency. This problem is solved as a result of the implementation of this system. Results in terms of accuracy include Decision Tree 0.9, Naive Bayes 0.99, Support Vector Machine 0.97, Logistic Regression 0.95, Random Forest 0.99, and XGBoost 0.99.