Pedosphere 20(3): 334--341, 2010
ISSN 1002-0160/CN 32-1315/P
©2010 Soil Science Society of China
Published by Elsevier B.V. and Science Press
| Soil quality assessment using weighted fuzzy association rules |
XUE Yue-Ju1, LIU Shu-Guang2,3, HU Yue-Ming 1 and YANG Jing-Feng1 |
1 South China Agricultural University, Guangzhou 510642 (China)
2 U.S. Geological Survey Center for Earth Resources Observation and Science, Sioux Falls, SD 57198 (USA)
3 Geographic Information Science Center of Excellence, South Dakota State University, Brookings, SD 57007 (USA) |
| ABSTRACT |
| Fuzzy association rules (FARs) can be powerful in assessing regional soil quality, a critical step prior to land planning and utilization; however, traditional FARs mined from soil quality database, ignoring the importance variability of the rules, can be redundant and far from optimal. In this study, we developed a method applying different weights to traditional FARs to improve accuracy of soil quality assessment. After the FARs for soil quality assessment were mined, redundant rules were eliminated according to whether the rules were significant or not in reducing the complexity of the soil quality assessment models and in improving the comprehensibility of FARs. The global weights, each representing the importance of a FAR in soil quality assessment, were then introduced and refined using a gradient descent optimization method. This method was applied to the assessment of soil resources conditions in Guangdong Province, China. The new approach had an accuracy of 87%, when 15 rules were mined, as compared with 76% from the traditional approach. The accuracy increased to 96% when 32 rules were mined, in contrast to 88% from the traditional approach. These results demonstrated an improved comprehensibility of FARs and a high accuracy of the proposed method. |
| Key Words: accuracy, comprehensibility, global weights |
| Citation: Xue, Y. J., Liu, S. G., Hu, Y. M. and Yang, J. F. 2010. Soil quality assessment using weighted fuzzy association rules. Pedosphere. 20(3): 334-341. |
| |
|
View Full Text
|
|
|
|
