Advanced search

Application of support vector machine (SVM) method to predict the distribution of Azolla filiculoides (Lam.) in Selkeh Wildlife Refuge, Anzali Wetland, northern Iran

Author
Organization
Abstract
Support vector machines (SVMs) were used to predict the distribution of Azolla filiculoides (Lam.) in Selkeh ecosystem (northern Iran). Various structural-habitat and physico-chemical variables were used as inputs for the development of SVMs. First SVMs were tested with different folds cross-validation and then they were optimised with different exponents in order to find the best predictive results as well as a reliable prediction for Azolla distribution. The biotic variable consisted of cover percentage of Azolla in terms of 3 classes (low, medium and high). The quality of models was assessed using the percentage of Correctly Classified Instances (CCI %) and Cohen's kappa statistics (k). The results of the present study demonstrated that SVMs yielded the high performances to predict Azolla in the sampling sites. By attribute weights, SVMs provided quantitative correlations between ecosystem characteristics and the distribution of Azolla and accordingly allowed a be tter ecological understanding of the obtained results. The structural-habitat variables played a key role for the prediction than water quality ones. The depth of wetland and dissolved oxygen were the two most important explanatory variables in the study area. SVMs hence proved to have high possibility for decision-making in ecosystem restoration and conservation management program.
Keywords
support vector machines, Ecological modelling, ecosystem management

Citation

Please use this url to cite or link to this publication:

Chicago
Sadeghi Pasvisheh, Roghayeh, Rahmat Zarkami, Karim Sabetraftar, and Patrick Van Damme. 2011. “Application of Support Vector Machine (SVM) Method to Predict the Distribution of Azolla Filiculoides (Lam.) in Selkeh Wildlife Refuge, Anzali Wetland, Northern Iran.” In Forests 2011, Abstracts.
APA
Sadeghi Pasvisheh, R., Zarkami, R., Sabetraftar, K., & Van Damme, P. (2011). Application of support vector machine (SVM) method to predict the distribution of Azolla filiculoides (Lam.) in Selkeh Wildlife Refuge, Anzali Wetland, northern Iran. Forests 2011, Abstracts. Presented at the Forests 2011 : Forests 2011 : Conservation and management of forests for sustainable development : where science meets policy.
Vancouver
1.
Sadeghi Pasvisheh R, Zarkami R, Sabetraftar K, Van Damme P. Application of support vector machine (SVM) method to predict the distribution of Azolla filiculoides (Lam.) in Selkeh Wildlife Refuge, Anzali Wetland, northern Iran. Forests 2011, Abstracts. 2011.
MLA
Sadeghi Pasvisheh, Roghayeh, Rahmat Zarkami, Karim Sabetraftar, et al. “Application of Support Vector Machine (SVM) Method to Predict the Distribution of Azolla Filiculoides (Lam.) in Selkeh Wildlife Refuge, Anzali Wetland, Northern Iran.” Forests 2011, Abstracts. 2011. Print.
@inproceedings{3039445,
  abstract     = {Support vector machines (SVMs) were used to predict the distribution of Azolla filiculoides (Lam.) in Selkeh ecosystem (northern Iran). Various structural-habitat and physico-chemical variables were used as inputs for the development of SVMs. First SVMs were tested with different folds cross-validation and then they were optimised with different exponents in order to find the best predictive results as well as a reliable prediction for Azolla distribution. The biotic variable consisted of cover percentage of Azolla in terms of 3 classes (low, medium and high). The quality of models was assessed using the percentage of Correctly Classified Instances (CCI \%) and Cohen's kappa statistics (k). The results of the present study demonstrated that SVMs yielded the high performances to predict Azolla in the sampling sites. By attribute weights, SVMs provided quantitative correlations between ecosystem characteristics and the distribution of Azolla and accordingly allowed a be tter ecological understanding of the obtained results. The structural-habitat variables played a key role for the prediction than water quality ones. The depth of wetland and dissolved oxygen were the two most important explanatory variables in the study area. SVMs hence proved to have high possibility for decision-making in ecosystem restoration and conservation management program.},
  author       = {Sadeghi Pasvisheh, Roghayeh and Zarkami, Rahmat and Sabetraftar, Karim and Van Damme, Patrick},
  booktitle    = {Forests 2011, Abstracts},
  keyword      = {support vector machines,Ecological modelling,ecosystem management},
  language     = {eng},
  location     = {Leuven, Belgium},
  title        = {Application of support vector machine (SVM) method to predict the distribution of Azolla filiculoides (Lam.) in Selkeh Wildlife Refuge, Anzali Wetland, northern Iran},
  year         = {2011},
}