Choosing the Best Method for Filling in Missing Data in Gene Expression Studies
Author Information
Author(s): Brock Guy N, Shaffer John R, Blakesley Richard E, Lotz Meredith J, Tseng George C
Primary Institution: University of Louisville and University of Pittsburgh
Hypothesis
Which missing value imputation method is optimal for different types of gene expression data?
Conclusion
The study found that LSA, LLS, and BPCA are the top-performing imputation methods, with no single method being superior across all data sets.
Supporting Evidence
- The study evaluated eight imputation methods including KNN, SVD, and BPCA.
- LSA, LLS, and BPCA were found to be the most competitive methods.
- The performance of imputation methods varied significantly based on data complexity.
Takeaway
This study looked at different ways to fill in missing data in gene studies and found that some methods work better depending on the type of data.
Methodology
The study evaluated eight imputation methods on nine different microarray data sets and proposed two selection schemes for optimal method selection.
Potential Biases
Potential bias in method selection due to the reliance on specific data sets for model fitting.
Limitations
The study does not cover all possible imputation methods and is based on a limited number of data sets.
Statistical Information
P-Value
p < 0.01
Statistical Significance
p < 0.01
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website