This study utilized the machine learning algorithm called "Bagging" to gain a better understanding of the relationships between climate variability, crop yields, and the agricultural contributions to the gross domestic product (agrigdp) in the Nile basin. The multiple stepwise regression model (MSRM) was also used as a comparison. The time series data collected from 1961 to 2016 for four main crops (groundnuts, sorghum, wheat, and sugarcane), rainfall, air temperature (minimum, mean, and maximum), and agrigdp were divided into 70% for training and 30% for testing using R packages. The results indicated that the Bagging algorithm improved crop yield predictions by 18%—63% compared to the MSRM. All crops demonstrated a greater sensitivity to climate variability, resulting in a low probability of achieving the highest attainable crop yield: 18% – 33% for groundnut, 18%—38% for sorghum, 21%—41% for wheat, and 22%—56% for sugarcane. The climate variability also led to a basin-wide loss of 35 million US$ in the agrigdp, with a basin-wide probability of 29%—61% for attaining the highest agrigdp. The algorithm clearly identified the ideal climatic conditions for reaching the highest attainable crop yields and agrigdp in the Nile Basin. The central factor was found to be air temperature, particularly the mean temperature, with anomalies of ± 0.5 °C responsible for 17%—69% of crop yield losses. Strategies for adapting to and managing risks associated with climate variability should focus on on-farm air temperature management practices, technologies such as greenhouses, and plant breeding.

这项研究利用称为“Bagging”的机器学习算法来更好地了解尼罗河流域气候变化、作物产量以及农业对国内生产总值 (agrigdp) 的贡献之间的关系。多元逐步回归模型（MSRM）也被用作比较。 1961年至2016年收集的四种主要作物（花生、高粱、小麦和甘蔗）、降雨量、气温（最小值、平均值和最大值）和agrigdp的时间序列数据，分为70%用于训练，30%用于训练使用 R 包进行测试。结果表明，与 MSRM 相比，Bagging 算法将作物产量预测提高了 18%—63%。所有作物都表现出对气候变化的更大敏感性，导致实现最高作物产量的可能性较低：花生为 18% - 33%，高粱为 18% - 38%，小麦为 21% - 41%，以及 22% —甘蔗为 56%。气候变化还导致全流域农业发展损失达3500万美元，全流域达到最高农业发展水平的概率为29%—61%。该算法清楚地确定了尼罗河流域实现最高作物产量和 agrigdp 的理想气候条件。研究发现，核心因素是气温，特别是平均温度，±0.5°C 的异常会造成 17%—69% 的作物产量损失。适应和管理与气候变化相关的风险的策略应侧重于农场气温管理实践、温室等技术和植物育种。