Crop Yield Prediction Using Machine Learning
M. Sunil Kumar, S. Girinath, G Lakshmi, Arjun Ganesh, Kapil Kumar
Abstract
We are all aware of India's reliance on agriculture. The user may forecast the agricultural production in any year they choose using the script's simple criteria, which include State, district, season, and region. Kernel Ridge, Lasso, and ENet algorithms are used in the research to estimate the yield, and the idea of stacking regression is used to enhance the algorithms and produce a more precise prediction. The script offers a user-friendly tool for forecasting agricultural yields in India by integrating various methodologies and taking into account straightforward criteria like state, district, season, and area. For farmers, decision-makers, and researchers who are interested in comprehending and improving agricultural productivity in the nation, this information may be useful. By taking into account elements like temperature, rainfall, acreage, and other parameters, the predictions provided by learning algorithms will assist farmers in choosing which crop to cultivate to provide the maximum yield. This links the agricultural and technology sectors together. The DNN strategy created by the winning team outperformed alternative strategies in the Syngenta Crop Challenge if it had a greater prediction accuracy than other competing models. Their model probably used cutting-edge methods for data preprocessing, feature engineering, and network architecture design, allowing it to more accurately represent the complex interactions between genotype, environment, and crop production.