Predictive Modeling of Agricultural Production Trends using Machine Learning: A Random Forest Approach

Abstract

Crop production forecasting is crucial for formulating strategies and allocating resource because it acts a crucial part in ensuring worldwide security of food. Utilising a collection of data that spans between 1961 to 2007, the present research compares predictive machine learning techniques for estimating crop outcomes across different nations. To make sure the collection of data, which included 311,624 items, could be used employing machine learning theories, it was preprocessed using the techniques of feature engineering and category encoding. We used the Random Forest Regressor and the Gradient Boosting Regressor, two sophisticated prediction models. Both the Random Forest Regressor with the Gradient Boosting Regressor, 2 powerful models for forecasting. Having an R-squared score of 1.00 & a Mean Squared Error (MSE) of 1.11*10^12, that implies nearly ideal accuracy in prediction, the Random Forest Regressor scored better than the other models. Excellent outcomes were achieved as well using the Gradient Boosting Regressor, but using slightly reduced precision measures.

Keywords: Artificial intelligence, Crop production, Gradient boosting regressor, Machine learning algorithm, Random forest regressor.

View PDF