基于K-Means聚类算法的地区农业差异性分类研究
Research on the Classification of Regional Agricultural Differences Based on K-Means Clustering Algorithm

农业是国民经济的基础，不仅为我们的生存和发展提供了基本的生活资料，还对社会经济的发展与稳定具有至关重要的作用。因此，从农业差异性的角度，对我国各地区进行分类，可以为制定更具针对性和科学性的政策提供数据支持，其具有重大意义。本文通过选取1999年至2019年我国31个地区的地区生产总值及农业总产值，建立了基于K-means聚类算法的地区农业差异性分类研究模型。首先，以农业经济占比率作为聚类基础，利用K-means聚类算法将31个地区分为农业为主型地区和非农业为主型地区；进一步地，为优化聚类结果，在K-means聚类算法中结合肘部法则，确定最优聚类数为4类，进而将31个地区分为农业高质量地区，农业中质量地区，农业发展一般地区及农业欠发展地区，且通过三个评价指标的优化率，即轮廓系数优化率为10.40%；DBI优化率为28.42%，CH优化率为104.72%，可以看出优化后的聚类效果得到了显著提升；最后，基于分类结果，对4类地区，提出与之相适应的农业发展建议，为相关地区在制定农业发展规划时提供针对性意见。
Agriculture is the foundation of the national economy. It not only provides us with the basic means of subsistence and development, but also plays a crucial role in the development and stability of the social economy. Therefore, classifying various regions in China from the perspective of agricultural differences can provide data support for formulating more targeted and scientific policies, which is of great significance. This paper selects the gross agricultural output value and regional gross domestic product of 31 regions in China from 1999 to 2019, and establishes a research model for classifying agricultural regions based on K-means clustering algorithm. Firstly, taking the proportion of agricultural economy as the basis of clustering, the K-means clustering algorithm is used to divide the 31 regions into regions mainly based on agriculture and regions not mainly based on agriculture. Furthermore, in order to optimize the clustering results, the elbow method is combined with the K-means algorithm, and the optimal number of clusters is determined to be 4. Then, the 31 regions are divided into four categories, namely regions with high-quality agriculture, regions with medium-quality agriculture, regions with general agricultural development, and regions with underdeveloped agriculture. And through the optimization rates of three evaluation indicators, that is, the silhouette coefficient optimization rate is 10.40%; the DBI optimization rate is 28.42%, and the CH optimization rate is 104.72%. It can be seen that the clustering effect after optimization has been significantly improved. Finally, the corresponding agricultural development suggestions are put forward for the four classified regions based on the classification results, which providing targeted opinions when formulating agricultural development plans.