India's agricultural productivity is lower than in some other countries. For example, wheat yields are about 2.7 tons of hectares in India compared to six tons in China.
While technical aids such as drones and sensors can help to enhance agricultural production, artificial intelligence (machine learning) can prove to be an even bigger game changer, especially when deciding on the next crop to grow for improved yields and profits.
The “ML-based crop recommendation system” is the next big thing in agriculture today. Farmers with over 145 million small farms in India, with mostly less than 1.1 hectares, need clear database guidance to select the right crops to improve income and resilience to climate change.
In this, two independent studies conclude that the prediction accuracy of the “random forest” ML model is the highest. The “Random Forest” model combines multiple “decision trees.” This is an ML algorithm that uses tree-like structures to make predictions.
The first work was by scientist Stephen Sam and Silima Marshall Dubleo of Brunel University in London. They looked at 12,389 data points of 19 crops in 15 Indian states between 2011 and 14.
“We will combine environmental and economic input parameters to develop and evaluate the accuracy of two machine learning models ('Random Forest' and 'Support Vector Machines') to recommend highly and profitable crops for farmers.”
They concluded that “random forests based on lag variables” (past values of data points used to predict the future) are the most accurate.
Various conditions
Researchers tested two computer-based models to see how well they could propose the right crop. One method showed high accuracy, but was not realistic as it did not take into account how the crop condition changes over time.
To balance accuracy with real-world utility, researchers introduced “lag variables” to improve the performance of the model. Ultimately, the model using the random forest method with a time-held approach worked best for crop recommendations in India.
The study emphasizes that examining both market and environmental factors will provide better advice for farmers. It is also suggested that future improvements should include data such as market demand, prices, returns, etc. to further adapt recommendations that are more suitable for India's diverse agricultural conditions.
Another research paper entitled “Crop Recommendation System Using Machine Learning” by researchers at Prakasam Engineering College in Khandukur, Andhra Pradesh, concluded that the random forest model is the best with an accuracy of 99.3%.
“The system recommends the best crop across 22 different crop categories, contributing to increased agricultural productivity and sustainable agricultural practices,” says Dr. Mlakshma Rao and his student Dr. Soprala Naveena.
“The crop recommendation system represents the success of the integration of machine learning technology and agricultural science, creating tools to bridge the gap between sophisticated analytical capabilities and practical agricultural applications,” they said, adding that the system “serves as evidence of the concepts of artificial intelligence and machine learning that support sustainable, productive and equitable agricultural systems.”
Something like this
Released on June 29, 2025
