Classification of SNPs for obesity analysis using FARNeM modelling
Recent research found that genetics plays an important role in obesity risk analysis besides life styles. Many literatures are focusing on analyzing the effect of Single Nucleotide Polymorphism (SNPs) towards obesity to facilitate personalized medication. However, SNPs data are normally large and no...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Conference or Workshop Item |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://eprints.utem.edu.my/id/eprint/11906/2/paper68.pdf http://eprints.utem.edu.my/id/eprint/11906/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Recent research found that genetics plays an important role in obesity risk analysis besides life styles. Many literatures are focusing on analyzing the effect of Single Nucleotide Polymorphism (SNPs) towards obesity to facilitate personalized medication. However, SNPs data are normally large and noisy, which affects the accuracy and computational complexity on data processing and analysis. Therefore, efficient data reduction is essential to yield better analysis results and reduce computational complexity in the experimentations. In this paper, we investigated feature selection process in obesity related SPNs analysis using Forward attribute reduction based on neighbourhood rough set model (FARNeM). The experimental results were compared against Correlation Feature Selection (CFS) method and ReliefF method. Classification accuracy, sensitivity, specificity, positive predictive value and negative predictive value were chosen to assess the performance of the comparison methods on error rate and validated by paired-sample T-test. FARNeM has outperformed other comparison techniques by having three highest performances which are specificity, positive predictive value and negative predictive value. But, FARNeM did not achieve good reduction rate when applied to the experimental data set. However, the overall analysis showed that, it is encouraging to include feature selection process before the learning algorithms. |
|---|