以某南方大型城市A区的2 394条地址数据为基础，引入混淆矩阵法对百度、高德、腾讯、天地图4个平台的精度指标进行定量分析，并利用聚类分析进行分类。实验结果表明，百度的总体编码数据质量最高；在各平台的指标中，高德的“门牌号”、百度的“80”、腾讯的“9”、天地图的“90”等指标表现稳定且具有较高精度，其中高德的“门牌号”数据质量最高，约70%的误差在50 m以内，超过90%的误差在500 m以内；其他指标的排序与网站介绍的等级顺序存在差异。研究结果可为网络地理编码平台用户选择数据提供依据。

Based on 2 394 address data in District A of a large southern city, we used confusion matrix method to quantitatively analyze the accuracy indexes of Baidu, AutoNavi, Tencent and MapWorld, and used cluster analysis for classification. The experimental results show that Baidu has the highest overall coding data quality. Among the indexes of various platforms,“house number”of AutoNavi,“80”of Baidu,“9”of Tencent, and“90”of MapWorld perform have stable performance and high accuracy. Among them,“house number”of AutoNavi has the highest quality, with about 70% of the error below 50 m, and more than 90% of the errors within 500 m. The ranking of other indexes differs from the rank order described on the site. The results of this study can provide a basis for the users of network geocoding platform to select data.