• Home
  • Research
  • Publications
  • Software
  • Students
  • Teaching
R packages
​
Visit my GitHub page here ​for a complete list of R packages.
  • ​DIMM: Performs distributed and integrated method of moments regression for high- dimensional correlated responses. Divides outcomes into blocks, analyses blocks using composite likelihood, and combines estimators using a one-step update or an optimal generalized method of moments (GMM) equation. See this paper for reference.
  • DDIMM: Performs doubly distributed and integrated method of moments regression for high-dimensional correlated responses. Divides outcomes and subjects into blocks, analyses blocks using composite likelihood or generalized estimating equations, and combines estimators using a one-step update or an optimal GMM equation. See this paper for reference.
  • ​DIQIF: Performs doubly distributed and integrated method of moments regression for high-dimensional correlated responses. Divides outcomes and subjects into blocks, analyses blocks using composite likelihood, generalized estimating equations or quadratic inference functions, and combines estimators using a one-step update or an optimal GMM equation following a pre-specified homogeneity partition index. See this paper for reference.
  • BRdac: Performs regression analysis of Brown-Resnick and inverted Brown-Resnick max-stable spatial data using a divide-and-conquer approach. Divides the spatial domain into blocks, analyses blocks using censored composite likelihood, and combines estimators using a one-step update. Extensions for spatially-varying coefficient models are included. See this paper for reference.
  • ​SLA: Estimates mean regression parameters for longitudinal data using an online streaming procedure. The longitudinal data are observed in sequential streams that are sequentially analyzed using a time-varying regression model. Correlation between outcomes is modeled using a first-order autoregressive process. See this paper for reference.
  • ISEDI: Estimates the parameters of a generalized linear model in a dataset by borrowing information from a prior analysis on another dataset. Estimation is based on an information-shrinkage estimator that introduces discounted units of information from the prior analysis into the current analysis. See this paper for reference.

Julia packages
  • RECaST: Estimates recalibration parameters in Cauchy random effect model for transfer learning from a source model to a target model. Includes code for comparisons to other transfer learning approaches. See this paper for reference.
Proudly powered by Weebly