Environmental Management

, Volume 34, Issue 4, pp 546–558 | Cite as

Rough Set Rule Induction for Suitability Assessment

Article

Abstract

The data that characterize an environmental system are a fundamental part of an environmental decision-support system. However, obtaining complete and consistent data sets for regional studies can be difficult. Data sets are often available only for small study areas within the region, whereas the data themselves contain uncertainty because of system complexity, differences in methodology, or data collection errors. This paper presents rough-set rule induction as one way to deal with data uncertainty while creating predictive if–then rules that generalize data values to the entire region. The approach is illustrated by determining the crop suitability of 14 crops for the agricultural soils of the Willamette River Basin, Oregon, USA. To implement this method, environmental and crop yield data were spatially related to individual soil units, forming the examples needed for the rule induction process. Next, four learning algorithms were defined by using different subsets of environmental attributes. ROSETTA, a software system for rough set analysis, was then used to generate rules using each algorithm. Cross-validation analysis showed that all crops had at least one algorithm with an accuracy rate greater than 68%. After selecting a preferred algorithm, the induced classifier was used to predict the crop suitability of each crop for the unclassified soils. The results suggest that rough set rule induction is a useful method for data generalization and suitability analysis.

Keywords

Crop suitability assessment Rule induction Rough set theory Regional modeling 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Agotnes, T. 1999. Filtering large propositional rule sets while retaining classifier performance. MSc thesis, Norwegian University of Science and Technology, Department of Computer and Information Science, February 1999.Google Scholar
  2. 2.
    Alpaydin, E. 1999Combined 5 × 2 CV F-Test for comparing supervised classification learning algorithms.Neural Computation1118851982CrossRefPubMedGoogle Scholar
  3. 3.
    An, A., Shan, N., Chan, C., Cercone, N., Ziarko, W. 1996Discovering rules from data for water demand prediction.Engineering Applications of Artificial Intelligence9645653CrossRefGoogle Scholar
  4. 4.
    Berger, P., Bolte, J. (2004) Evaluating the impact of policy options on agricultural landscapes: an alternative futures approach. Ecological Applications. 14: 342–354Google Scholar
  5. 5.
    Bruha, I. 1997

    Quality of decision rules: definitions and classification schemes for multiple rules. Pages 107–131

    Nakhaeizadeh, G.Taylor, C. C. eds. Machine learning and statistics, the interface.John Wiley and SonsNew York
    Google Scholar
  6. 6.
    Daly, C., Neilson, R. P., Phillips, D. L. 1994A statistical-topographic model for mapping climatological precipitation over mountainous terrain.Journal of Applied Meteorology33140158CrossRefGoogle Scholar
  7. 7.
    Dietterich, T. G. 1998Approximate statistical tests for comparing supervised classification learning algorithms.Neural Computation1018951923CrossRefPubMedGoogle Scholar
  8. 8.
    Dougherty, J., Kohavi, R., Sahami, M. 1995

    Supervised and unsupervised discretizations of continuous features. Pages 194–202

    Prieditis, A.Russell, S. eds. Proceedings of the 12th international conference on machine learning (ML95)Morgan KaufmannSan Francisco
    Google Scholar
  9. 9.
    Furuta, H., Hirokane, M., Mikumo, Y. 1998

    Extraction method based on rough set theory of rule-type knowledge from diagnostic cases of slope-failure danger levels. Pages 178–192

    Polkowski, L.Skowron, A. eds. Rough sets in knowledge discovery 2: applications, case studies and software systems.Physica-VerlagHeidelberg
    Google Scholar
  10. 10.
    Jenssen, T. K. 1998. Refinements to Mollestad’s algorithm for synthesis of default rules. MSc thesis, Norwegian University of Science and Technology, Department of Computer and Information Science.Google Scholar
  11. 11.
    Johnson, D. S. 1974Approximation algorithms for combinatorial problems.Journal of Computer and System Sciences9256278Google Scholar
  12. 12.
    Komorowski, J., Pawlak, Z., Polkowski, L., Skowron, A. 1999

    Rough sets: a tutorial. Pages 3–98

    Pal, S. K.Skowron, A. eds. Rough-fuzzy hybridization: a new trend in decision making.Springer-VerlagSingapore
    Google Scholar
  13. 13.
    Øhrn, A., Komorowski, J., Skowron, A., Synak, P. 1998

    The ROSETTA software system

    Polkowski, L.Skowron, A. eds. Rough sets in knowledge discovery 2: applications, case studies and software systems.Physica-VerlagHeidelberg572576
    Google Scholar
  14. 14.
    Omernik, J. M. 1987Ecoregions of the conterminous United States:Annals of the Association of American Geographers771181251 pl., scale 1:7,500,000CrossRefGoogle Scholar
  15. 15.
    Pater, D. E., Bryce, S. A., Thorson, T. D., Kagan, J., Chappell, C.,Omernik, J. M., Azevedo, S. H., Woods, A. J. 1998. Ecoregions of Western Washington and Oregon (two-sided color poster with map, descriptive text, summary tables, and photographs). U.S. Geological Survey, Reston, VA.Google Scholar
  16. 16.
    Pawlak, Z. 1982Rough sets.International Journal of Computer and Information Sciences11341356CrossRefGoogle Scholar
  17. 17.
    Pawlak, Z. 1984Rough classification.International Journal of Man-Machine Studies20469483Google Scholar
  18. 18.
    Pawlak, Z. 1991Rough sets: theoretical aspects of reasoning about data.Kluwer Academic PublishersDordrecht229Google Scholar
  19. 19.
    Soil Survey Staff. 2001. National Soil Survey Handbook, title 430-VI. Natural Resources Conservation Service.Google Scholar
  20. 20.
    U. S. Department of Agriculture, Natural Resources Conservation Service. 1998. Soil Survey Geographic (SSURGO) Database. Fort Worth, Texas.Google Scholar
  21. 21.
    U.S. Environmental Protection Agency. 1996. Level III ecoregions of the continental United States (revision of Omernik 1987): Corvallis, OR, U. S. Environmental Protection Agency, digital map, scale 1:250,000.Google Scholar
  22. 22.
    van Diepen, C. A., van Keulen, H., Wolf, J., Berkhout, J. A. A. 1991

    Land evaluation: from intuition to quantification. Pages 139–204

    Stewart, B. A. eds. Advances in soil science.Springer-VerlagNew York
    Google Scholar
  23. 23.
    Wang, F. 1994The use of artificial neural networks in a geographical information system for agricultural land-suitability assessment.Environment and Planning A26265284Google Scholar
  24. 24.
    Wilk, S., Flinkman, M., Michalowski, W., Nilsson, S., Slowinski, R., Susmaga, R. 1998. Identification of biodiversity and other forest attributes for sustainable forest management: Siberian Forest case study. IIASA Interim Report 98–106. International Institute for Applied Systems Analysis. Laxenburg, Austria, 23 pp.Google Scholar
  25. 25.
    Withrow-Robinson, B., Hibbs, D., Beuter, J. 1995. Poplar chip production for Willamette Valley grass seed sites. Forest Research Laboratory, Oregon State University, Corvallis, OR, 47 pp.Google Scholar

Copyright information

© Springer-Verlag New York, Inc. 2004

Authors and Affiliations

  1. 1.Department of BioengineeringOregon State UniversityCorvallisUSA

Personalised recommendations