Search ETDs:
A Comparison of Rule Extraction Techniques with Emphasis on Heuristics for Imbalanced Datasets
Singh, Manjeet

2010, Master of Science (MS), Ohio University, Industrial and Systems Engineering (Engineering and Technology).
Ecological datasets are not analyzed accurately due to the presence of imbalance and outliers. Imbalance being the major cause for unacceptable modeling results obtained in many cases. This research provides a solution with acceptable modeling accuracy values. SMOTE (in a modified form) will be used to preprocess the data. Different techniques such as traditional techniques (Regression and Stepwise Regression), Artificial Neural Networks and TREPAN will be used in conjunction with SMOTE to find out the combination which gives the best results. 3D Surfaces will be generated to decipher the impact of different inputs on the output and the interaction between inputs at a set output value.
Gary Weckman, PhD (Advisor)
Namkyu Park, PhD (Committee Member)
Tao Yuan, PhD (Committee Member)
David Millie, PhD (Committee Member)
110 p.

Recommended Citations

Hide/Show APA Citation

Singh, M. (2010). A Comparison of Rule Extraction Techniques with Emphasis on Heuristics for Imbalanced Datasets. (Electronic Thesis or Dissertation). Retrieved from https://etd.ohiolink.edu/

Hide/Show MLA Citation

Singh, Manjeet. "A Comparison of Rule Extraction Techniques with Emphasis on Heuristics for Imbalanced Datasets." Electronic Thesis or Dissertation. Ohio University, 2010. OhioLINK Electronic Theses and Dissertations Center. 11 Dec 2017.

Hide/Show Chicago Citation

Singh, Manjeet "A Comparison of Rule Extraction Techniques with Emphasis on Heuristics for Imbalanced Datasets." Electronic Thesis or Dissertation. Ohio University, 2010. https://etd.ohiolink.edu/

Files

ohiou1282139633.pdf (2.32 MB) View|Download