Robust Clustering in High Dimensional Data Using Statistical Depths
Author Information
Author(s): Ding Yuanyuan, Dang Xin, Peng Hanxiang, Wilkins Dawn
Primary Institution: The University of Mississippi
Hypothesis
A new algorithm that is robust and works well in high dimensional data sets is needed.
Conclusion
The proposed clustering algorithm outperforms the componentwise-median-based bisecting k-median algorithm for high dimension and low sample size data.
Supporting Evidence
- The proposed algorithm outperformed the bisecting k-median algorithm in simulations.
- The spatial median is more robust in high dimensional data compared to the componentwise median.
- Results showed that the bisecting k-spatialMedian algorithm successfully identified clusters in noisy data.
Takeaway
This study created a new way to group data that works better when there are lots of features but not many examples.
Methodology
The study proposes a new robust divisive clustering algorithm based on statistical spatial depth and tests it on real gene expression data sets.
Limitations
The algorithm's performance may vary with different data sets and noise levels.
Participant Demographics
The study involved gene expression data from 62 samples, including 40 tumor and 22 normal colon tissues.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website