基于关联规则的耕地质量评价数据检错方法研究 ——以广州市为例

研究目的:从数据项之间关联关系的角度切入,探索一种新的耕地数据质量检错方法,以期更有效地提高耕地数据库的质量。研究方法:通过数据挖掘算法寻找耕地数据库中的关联关系,计算这些关联关系的发生频率,从中提取低频发生的关联关系作为检测规则(关联规则),最后利用这些关联规则识别耕地数据库中的错误记录(包含或符合关联规则的耕地数据记录为错误记录)。研究结果:(1)该方法有能力识别耕地数据库中的错误,可以做到有效提高耕地参评数据库的正确性;(2)经计算,与耕地领域现有的传统数据检错方法相比,同等条件下该方法可将检错效率提高11倍,甚至更多;(3)该方法可以针对不同的数据库迅速挖掘关联规则,灵活地应对不同的耕地数据库和层出不穷的错误类型。研究结论:基于关联规则的耕地数据库质量检测方法高效、便捷,为耕地领域现有的数据检错方法开辟了一个新的角度和思路,可以在地学领域广泛应用。

Research on Associated Rule-Based Error Checking Method on Assessment Index Database of Cultivated Land Quality: A Case Study on Guangzhou City

The purposes of this paper are to explore a new method of data quality checking of cultivated land data from the perspective of associated relationship between data items to improve the quality of cultivated land assessment index database more effectively. The research method of this paper is to find the associated relationships in the cultivated land database by data mining and calculate the frequency of occurrence of these associations. The low-frequency associations are extracted and will be used as the checking rules (associated rules) to identify the errors in the database. The results show that: 1)this method can find the vast majority of errors in the cultivated land database, and it can improve the accuracy of the cultivated land assessment index database effectively. 2)Through the calculation, the error checking efficiency of this method can be increased by 11 times or more under the same conditions, compared with the existing traditional manual error checking method in the field of cultivated land. 3)This method can promptly discern mining associated rules for different databases, and flexibly check different cultivated land databases and various types of errors. In conclusion, the cultivated land data quality checking method introduced in this paper is efficient and convenient, and provides a new perspective for the existing methods of data checking in the field of cultivated land, which is worthy of being widely used in the field of geosciences.