Ordinal classification in high-dimensions

Results from National Institutes of Health/National Institutes of Library Medicine funded projects
R03-LM009347-01A2 and R03-LM009347-02S1.
Kellie J. Archer, PI

R03-LM009347-01A2

Aim 1: Extend the recursive partitioning and random forest classification methodologies for predicting an ordinal response by developing computational tools for the R programming environment including implementing the ordinal impurity criteria in rpart and implementing the ordinal impurity criteria in randomForest

• Link to rpartOrdinal package available from the Comprehensive R Archive Network
• Archer KJ. rpartOrdinal: An R Package for Deriving a Classification Tree for Predicting an Ordinal Response, Journal of Statistical Software, 34(7):1--17.

Aim 2: Evaluate the proposed ordinal classification methods in comparison to existing nominal and continuous response methods using simulated, benchmark, and gene expression datasets.

• Archer KJ and Mas VR. Ordinal response prediction using bootstrap aggregation, with application to a high-throughput methylation dataset. Statistics in Medicine, Dec 20;28(29):3597-610, 2009.

Aim 3: Develop and evaluate methods for assessing variable importance when interest is in predicting an ordinal response.

• Archer KJ. “Variable Selection in High-dimensional Ordinal Class Prediction Problems with Genomics Applications.” Institute for Operations Research and the Management Sciences (INFORMS) Annual Meeting, Washington, D.C., October 13, 2008
• Archer KJ. “Variable selection for ordinal models with applications to high-dimensional data.” Eastern North American Region of the International Biometrics Society Annual Meeting, San Antonio, Texas, on March 18, 2009.
• Asomaning N, Archer KJ. High-throughput DNA methylation datasets for evaluating false discovery rate methodologies. Computational Statistics and Data Analysis, 56(6):1748-1756, 2012 [Epub ahead of print].

R03-LM009347-02S1 ARRA Competitive Supplement

Aim 1: Extend the L 1 penalized methodology to enable predicting an ordinal response by developing computational tools for the R programming environment.

• Link to glmnetcr package available from the Comprehensive R Archive Network
• Vignette of glmnetcr package
• Link to glmpathcr package available from the Comprehensive R Archive Network
• Overview of glmpathcr package

Aim 2: Using simulated, benchmark, and gene expression datasets, evaluate L1 penalized ordinal response models by comparing error rates from our L1 fitting algorithm to those obtained when using a forward variable selection modeling strategy and our ordinal random forest approach.

• Andre A.A. Williams and Kellie J. Archer. Stereotype logit models for high-dimensional data. Presented March 22, 2010 at the Eastern North American Region of the International Biometric Society, New Orleans, LA.
• Kellie J. Archer and Andre A.A. Williams. Penalized models for ordinal response prediction: Application discriminating patients with early stage Parkinson's disease. Presented March 23, 2010 at the Eastern North American Region of the International Biometric Society, New Orleans, LA.
• Kellie J. Archer, Valeria R. Mas, Daniel G. Maluf, Robert A. Fisher. Genes progressively up- or down-regulated across pre-neoplastic and neoplastic liver disease states. Presented May 1, 2010 at the American Transplant Congress, San Diego, CA.
• Andre A.A. Williams and Kellie J. Archer. Stereotype logit models for high-dimensional data. Presented August, 2010 at the Joint Statistical Meetings, Vancouver, British Columbia.

Aim 3: Evaluate methods for assessing important covariates from L1 penalized ordinal response models.

• Archer KJ, Williams AAA. L1 penalized continuation ratio models for ordinal response prediction using high-dimensional datasets. Statistics in Medicine, Feb 23. doi: 10.1002/sim.4484. [Epub ahead of print] 2012.
• Kellie J. Archer and Andre A.A. Williams. A Comparison of Frequentist and Bayesian Penalized Continuation Ratio Models for Predicting an Ordinal Response in High-Dimensional Datasets. Presented March 22, 2011 at the Eastern North American Region of the International Biometric Society, Miami, FL.

R03-LM009347-02S2 ARRA Administrative Supplement

GEO datasets

GEO Ordinal Outcome Datasets (xls)

GEO Time Series Datasets (xls)

GEO Dose Response Datasets (xls)

Home

Research

Publications

Research Support

R03-LM009347

R01-LM011169

Data and Software

Teaching

Service

Links

Contact

VCU Biostatistics