A Machine Learning Framework for Rainfall Prediction and Crop Recommendation in Precision Agriculture

SDGs: Primary SDG: SDG 2 - Zero Hunger. | Secondary SDGs: SDG 9 - Industry, Innovation and Infrastructure; SDG 12 - Responsible Consumption and Production; SDG 13 - Climate Action

Abstract

Rainfall is erratic, and farmers are unable to choose suitable crops. This has resulted in reduced agricultural output and losses in income. In the proposed system, Information about rainfall prediction and suggesting the appropriate crops in a scientific way is provided to the farmers. Studies show that rainfall predictions by machine learning model will be more accurate. Similarly, crop suggestion systems based on soil conditions will help to increase the productivity of the crop by taking into account the nutrition, climatic factors etc. On the basis of these ideas, the proposed system can predict the rainfall with the help of statistical and machine learning techniques. The suggest crops will be based on soil type (sandy, clay, loam), temperature, humidity, nitrogen level and the previous rainfall. The methodology is developed to be tested with readily available as well as area specific data and hence will be feasible. The prototype is developed in Python using Scikit-learn and the system will have web interface. It attains an accuracy of 96.82% in the crop recommendation, making it a tool for sustainable agriculture absolutely reliable, functional and easy to use. In addition, the use of a Random Forest model ensures transparency of the decisions, which is of vital importance in precision farming.