Impact of Data Normalization on Fuzzy Clustering of DNA Microarray Data
Author Information
Author(s): Kim Seo Young, Lee Jae Won, Bae Jong Sung
Primary Institution: Chonnam National University
Hypothesis
The study investigates how different data normalization methods affect the performance of fuzzy clustering algorithms on microarray datasets.
Conclusion
Lowess normalization is more effective for clustering genes from microarray data compared to other normalization methods, especially in noisy datasets.
Supporting Evidence
- Fuzzy C-Means clustering was shown to outperform hard clustering methods in noisy datasets.
- The optimal fuzzification parameter for FCM clustering was found to depend on the normalization method used.
- Lowess normalization provided higher membership values for genes compared to other normalization methods.
Takeaway
This study shows that how we prepare our data can change the results when grouping genes, and using the right method helps us find patterns better.
Methodology
The study applied Fuzzy C-Means clustering to three microarray datasets and three simulated datasets, comparing the effects of three normalization methods.
Limitations
The study primarily focused on three normalization methods and may not cover all possible approaches.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website