Working papers

• On pricing of rides on transportation networks (2023+). 

• Fast Bayesian structure learning in high-dimensional sparse Gaussian graphical models using decomposable covering (with A Thomas, F Rios).

Publications

• Parallel sampling of decomposable graphs using Markov chains on junction trees (to appear).  The Journal of American Statistical Association, Theory & Methods (2024). [Ar][Code]

• Fast global convergence of gradient descent for low-rank matrix approximation (with  Hengchao Chen, Xin Chen, and Qiang Sun). The Conference on Uncertainty in Artificial Intelligence (UAI) (2024). [Ar]

• Predictive inference for travel time on transportation networks (with  A Labbe, D Larocque and L Charlin).  The Annals of Applied Statistics 17 (4), 2796-2820 (2023). [Ar] [Code] 

• Using phylogeographic link‐prediction in primates to prioritize human parasite screening (with C S Werner, K Kasan, J K Geyer, M J Farrell and C L Nunn). American Journal of Biological Anthropology, (2020) 1– 12. [P]

• Predicting missing links in global host–parasite networks (with M Farrell, D Stephens and  J Davies). Journal of Animal Ecology 91 (4), (2022) 715-726. [P]

• A hierarchical Bayesian model for predicting ecological interactions using scaled evolutionary relationships (with M Farrell, D Stephens and  J Davies). The Annals of Applied Statistics 14 (1), (2020) 221-240. [Ar][P][Code]

• A Skew‐normal copula‐driven GLMM (with K Das and A Sen). Statistica Neerlandica 70 (4), (2016) 396-413. [Ar][P][Code]

parallelDG a Python (2.7) package that implements Bayesian inference for decomposable graphical models using Parallel Metropolis-Hastings. To install, in shell call pip install parallelDG

traveltimeCLT a R package that implements two methods for prediction of average travel time on a route and its uncertainty (variance). Data accompanies the package. To install, in R call devtools::install_github("melmasri/traveltimeCLT")

traveltimeHMM a R package that implements a Hidden Markov Model with a random trip effect to estimate the distribution of travel time. Data accompanies the package. To install, in R call devtools::install_github("melmasri/traveltimeHMM")

HPprediction  a R package that fits a hierarchical Bayesian model for predicting ecological interactions using scaled evolutionary relationships.  To install, in R call devtools::install_github("melmasri/HPprediction")

• Beyond L2 Loss - How we experiment with loss functions at Lyft.  Lyft Engineering (2019).

• On decomposable random graphs and link prediction models (PhD thesis, McGill University) 2017. Written under the supervision of David Stephens.

• A Skew-Normal Copula-Driven Generalized Linear Mixed Model for Longitudinal Data (MSc thesis, Concordia University) 2012. Written under the supervision of Arusharka Sen. 

• A Model to Decompose Inclusions in Educational Attainment Tables.
  UNESCO Institute for Statistics. Internal report. August 2014.
  Technical report produced as a consultant for the Methodology Unit of the UIS in 2014.

• A Model to Forecast Mean Years of Schooling Estimates.
  UNESCO Institute for Statistics. Internal report. August 2014.
  Technical report produced as a consultant for the Methodology Unit of the UIS in 2014.

• UIS Methodology For The Estimation of Mean Years of Schooling.
  UNESCO Institute for Statistics. December 2013.
  Joint work with Friedrich Huebler and Brenda Tay-Lim. [PDF].