Works in progress

Updated: 01 Aug 2024

R packages in development

msmd - An R package for fitting discrete-time Markov models to panel (longitudinal) data with time-varying covariates. Transition rates are parameterized by multinomial logistic regression models. Best-fit regression coefficients are estimated by numerical optimization, with asymptotic standard errors calcualted from the Hessian matrix. Estimated confidence intervals on state occupancy vs time curves are generated by resampling coefficients from the covariance matrix.

coverDAG - An R package for generating all exact covers of a set, subject to inclusion criteria as defined by a directed acyclic graph (DAG).

Collaborative work

Analysis of longitudinal student retention data - with Anna Bargagliotti, Cassandra Guarino, and Yiwang Li. Modelling student retention in undergraduate computer science programs using a competitive-risks survival analysis approach. Data is for ~3000 students at four campuses over ~10 years. Data analysis and visualization completed in R and STATA, using both multinomial logistic regression and markov model techniques.

Enumerating set covers subject to DAG-encoded contraints - with Anna Bargagliotti. Implemented in R, this algorithm revisits the classic set cover problem and considers the case where the underlying subsets are related to each other as nodes of a directed acyclic graph. Exact covers are valid only if they satisfy an inclusion constraint determined by the DAG: For subsets \(x\) and \(y\), if \(x \rightarrow y\) is an edge in the DAG, then any cover containing \(x\) must also contain \(y\).

Sampling compactness scores in a gerrymandering model - with LMU alumnus Joshua Mariz (’21). An algorithm written in Javascript to estimate the distribution of “compactness scores” in a grid model of electoral voting districts. The algorithm randomly generates districts of equal size in an n x n grid using a novel square-by-square growth method. Manuscript in preparation.

Markov model composition of Balinese reyong norot improvisations - with LMU alumnus Taylor Flanagan (’21). A Python-based computational project using Markov models to randomly generate compositions of Balinese gamelan gong kebyar improvisations on the reyong. Each of the reyong’s four players can play only some of the gamelan’s five tones and must use specific patterns. The model’s probability values come from a combination of top-down and bottom-up techniques, making extensive use of Leslie Tilley’s work on the grammar of reyong norot and example patterns from her concurrent study of musician Dewa Ketut Alit’s improvisation. The model outputs MIDI files for audio playback of the constructed songs. Manuscript in preparation.