Assessing Crop Yield Sensitivity to Climate Variables with Regression Models

Global agriculture must strike a balance between crop yield and the consequences of climate change, which affects important variables including temperature, precipitation, and CO2 emissions. A dataset covering ten countries—including China, India, Canada, and the United States—from 1990 to 2024 is used in this study to examine these interactions. A comprehensive analysis is ensured by the data's 30:70 test-to-training ratio, which covers important variables including agricultural output, soil health index, irrigation availability, pesticide and fertilizer use, and extreme weather occurrences. Regression models and time-series analysis were among the statistical and machine learning methods used to assess the correlations between these variables and forecast future trends. In comparison to the multiple linear regression model, which had an MSE of 0.9271, R2 of 0.0786, and MAE of 0.773, the linearly regression model outperformed with an MSE of 0.4679, R2 of 0.5349 and MAE of 0.5494. With lower MSE and MAE values signifying superior model performance, these findings demonstrate how well linear regression captures the intricate relationships between climate and agricultural variables. This study highlights the significance of adaptation techniques that have been demonstrated to increase yields and lessen the negative consequences of climate change, such as crop rotation and water management.