KRKellin N. Rumsey

Kellin N. Rumsey

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Kellin N. Rumsey has a total of 29 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • khaos: Bayesian Sparse Polynomial Chaos Expansion

    by Kellin Rumsey

    Implements Bayesian polynomial chaos expansion methods for surrogate modeling, uncertainty quantification, and sensitivity analysis. Includes sparse regression-based approaches and adaptive Bayesian models based on reversible-jump Markov chain Monte Carlo. Optional screening and basis-enrichment strategies are provided to improve scalability in moderate to high dimensions.

  • Open Access

    Bayesian Adaptive Polynomial Chaos Expansions

    by Kellin N. Rumsey, Devin Francom, Graham C. Gibson, J. Derek Tucker, Gabriel Huerta

    Polynomial chaos expansions (PCE) are widely used for uncertainty quantification (UQ) tasks, particularly in the applied mathematics community. However, PCE has received comparatively less attention in the statistics literature, and fully Bayesian formulations remain rare, especially with implementations in R. Motivated by the success of adaptive Bayesian machine learning models such as BART, BASS, and BPPR, we develop a new fully Bayesian adaptive PCE method with an efficient and accessible R implementation: khaos. Our approach includes a novel proposal distribution that enables data-driven interaction selection, and supports a modified g-prior tailored to PCE structure. Through simulation studies and real-world UQ applications, we demonstrate that Bayesian adaptive PCE provides competitive performance for surrogate modeling, global sensitivity analysis, and ordinal regression tasks.

  • Co-Active Subspace Methods for the Joint Analysis of Adjacent Computer Models

    by Kellin N. Rumsey, Zachary K. Hardy, Cory Ahrens, Scott Vander Wiel


  • Open Access

    Magnesium Oxide Reduces Anxiety-like Behavior in Mice by Inhibiting Sulfate-Reducing Bacteria

    by Cristina N. Coffman, Amanda Carroll-Portillo, Joe Alcock, Sudha B. Singh, Kellin Rumsey, Cody A. Braun, Bingye Xue, Henry C. Lin

    The gut microbiota–brain axis allows for bidirectional communication between the microbes in our gastrointestinal (GI) tract and the central nervous system. Psychological stress has been known to disrupt the gut microbiome (dysbiosis) leading to anxiety-like behavior. Pathogens administered into the gut have been reported to cause anxiety. Whether commensal bacteria affect the gut–brain axis is not well understood. In this study, we examined the impact of a commensal sulfate-reducing bacteria (SRB) and its metabolite, hydrogen sulfide (H2S), on anxiety-like behavior. We found that mice gavaged with SRB had increased anxiety-like behavior as measured by the open field test. We also tested the effects of magnesium oxide (MgO) on SRB growth both in vitro and in vivo using a water avoidance stress (WAS) model. We found that MgO inhibited SRB growth and H2S production in a dose-dependent fashion. Mice that underwent psychological stress using the WAS model were observed to have an overgrowth (bloom) of SRB (Deferribacterota) and increased anxiety-like behavior. However, WAS-induced overgrowth of SRB and anxiety-like behavioral effects were attenuated in animals fed a MgO-enriched diet. These findings supported a potential MgO-reversible relationship between WAS-induced SRB blooms and anxiety-like behavior.


  • Bayesian projection pursuit regression

    by Gavin Collins, Devin Francom, Kellin Rumsey


  • Open Access

    Discovering Active Subspaces for High-Dimensional Computer Models

    by Kellin N. Rumsey, Devin Francom, Scott Vander Wiel

    Dimension reduction techniques have long been an important topic in statistics, and active subspaces (AS) have received much attention this past decade in the computer experiments literature. The most common approach towards estimating the AS is to use Monte Carlo with numerical gradient evaluation. While sensible in some settings, this approach has obvious drawbacks. Recent research has demonstrated that active subspace calculations can be obtained in closed form, conditional on a Gaussian process (GP) surrogate, which can be limiting in high-dimensional settings for computational reasons. In this paper, we produce the relevant calculations for a more general case when the model of interest is a linear combination of tensor products. These general equations can be applied to the GP, recovering previous results as a special case, or applied to the models constructed by other regression techniques including multivariate adaptive regression splines (MARS). Using a MARS surrogate has many advantages including improved scaling, better estimation of active subspaces in high dimensions and the ability to handle a large number of prior distributions in closed form. In one real-world example, we obtain the active subspace of a radiation-transport code with 240 inputs and 9,372 model runs in under half an hour.

  • Open Access

    Generalized Bayesian MARS: Tools for Emulating Stochastic Computer Models

    by Kellin Rumsey, Devin Francom, Andy Shen

    The multivariate adaptive regression spline (MARS) approach of Friedman (1991) and its Bayesian counterpart (Francom et al. 2018) are effective approaches for the emulation of computer models. The traditional assumption of Gaussian errors limits the usefulness of MARS, and many popular alternatives, when dealing with stochastic computer models. We propose a generalized Bayesian MARS (GBMARS) framework which admits the broad class of generalized hyperbolic distributions as the induced likelihood function. This allows us to develop tools for the emulation of stochastic simulators which are parsimonious, scalable, interpretable and require minimal tuning, while providing powerful predictive and uncertainty quantification capabilities. GBMARS is capable of robust regression with t distributions, quantile regression with asymmetric Laplace distributions and a general form of "Normal-Wald" regression in which the shape of the error distribution and the structure of the mean function are learned simultaneously. We demonstrate the effectiveness of GBMARS on various stochastic computer models and we show that it compares favorably to several popular alternatives.

  • Open Access

    Understanding questions and concerns about potable water reuse: An analysis of survey write‐in responses

    by Caroline E. Scruggs, Catherine M. Heyne, Kellin N. Rumsey

    Urban centers around the world are grappling with the challenges associated with population increases, drought, and projected water shortages. Potable water reuse (i.e., purification of municipal wastewater for reuse as drinking water) is an option for supplementing existing water supplies. Public perception research on potable water reuse has predominantly employed surveys with multiple‐choice questions that constrain survey respondents to describe their concerns by choosing from several response options. This research examines hundreds of write‐in responses to a large public survey in Albuquerque, New Mexico, to provide a detailed analysis of residents' questions and concerns about potable water reuse. Findings demonstrate that allowing respondents to voice their actual concerns adds richness and nuance that cannot be obtained from multiple‐choice response data alone. Especially with controversial resource considerations, such as potable water reuse, planners would benefit from a full understanding of the problem before engaging with the community.


  • Open Access

    Mucin and Agitation Shape Predation of Escherichia coli by Lytic Coliphage

    by Amanda Carroll-Portillo, Kellin N. Rumsey, Cody A. Braun, Derek M. Lin, Cristina N. Coffman, Joe A. Alcock, Sudha B. Singh, Henry C. Lin

    The ability of bacteriophage (phage), abundant within the gastrointestinal microbiome, to regulate bacterial populations within the same micro-environment offers prophylactic and therapeutic opportunities. Bacteria and phage have both been shown to interact intimately with mucin, and these interactions invariably effect the outcomes of phage predation within the intestine. To better understand the influence of the gastrointestinal micro-environment on phage predation, we employed enclosed, in vitro systems to investigate the roles of mucin concentration and agitation as a function of phage type and number on bacterial killing. Using two lytic coliphage, T4 and PhiX174, bacterial viability was quantified following exposure to phages at different multiplicities of infection (MOI) within increasing, physiological levels of mucin (0–4%) with and without agitation. Comparison of bacterial viability outcomes demonstrated that at low MOI, agitation in combination with higher mucin concentration (>2%) inhibited phage predation by both phages. However, when MOI was increased, PhiX predation was recovered regardless of mucin concentration or agitation. In contrast, only constant agitation of samples containing a high MOI of T4 demonstrated phage predation; briefly agitated samples remained hindered. Our results demonstrate that each phage–bacteria pairing is uniquely influenced by environmental factors, and these should be considered when determining the potential efficacy of phage predation under homeostatic or therapeutic circumstances.