skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


Title: A Relaxation Approach to Feature Selection for Linear Mixed Effects Models
Linear Mixed-Effects (LME) models are a fundamental tool for modeling correlated data, including cohort studies, longitudinal data analysis, and meta-analysis. Design and analysis of variable selection methods for LMEs is more difficult than for linear regression because LME models are nonlinear. In this article we propose a novel optimization strategy that enables a wide range of variable selection methods for LMEs using both convex and nonconvex regularizers, including 𝓁1, Adaptive-𝓁1, SCAD, and 𝓁0. The computational framework only requires the proximal operator for each regularizer to be readily computable, and the implementation is available in an open source python package pysr3, consistent with the sklearn standard. The numerical results on simulated data sets indicate that the proposed strategy improves on the state of the art for both accuracy and compute time. The variable selection techniques are also validated on a real example using a data set on bullying victimization. Supplementary materials for this article are available online.  more » « less
Award ID(s):
1908890
PAR ID:
10470298
Author(s) / Creator(s):
; ; ; ;
Corporate Creator(s):
Editor(s):
Jones, G.; Faming, L.
Publisher / Repository:
Taylor & Francis Online
Date Published:
Journal Name:
Journal of Computational and Graphical Statistics
ISSN:
1061-8600
Page Range / eLocation ID:
1 to 42
Subject(s) / Keyword(s):
feature selection, mixed effects models, nonconvex optimization
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; ; ; ; (, Scientific Data)
    Abstract Mapped monthly data products of surface ocean acidification indicators from 1998 to 2022 on a 0.25° by 0.25° spatial grid have been developed for eleven U.S. large marine ecosystems (LMEs). The data products were constructed using observations from the Surface Ocean CO2Atlas, co-located surface ocean properties, and two types of machine learning algorithms: Gaussian mixture models to organize LMEs into clusters of similar environmental variability and random forest regressions (RFRs) that were trained and applied within each cluster to spatiotemporally interpolate the observational data. The data products, called RFR-LMEs, have been averaged into regional timeseries to summarize the status of ocean acidification in U.S. coastal waters, showing a domain-wide carbon dioxide partial pressure increase of 1.4 ± 0.4 ÎŒatm yr−1and pH decrease of 0.0014 ± 0.0004 yr−1. RFR-LMEs have been evaluated via comparisons to discrete shipboard data, fixed timeseries, and other mapped surface ocean carbon chemistry data products. Regionally averaged timeseries of RFR-LME indicators are provided online through the NOAA National Marine Ecosystem Status web portal. 
    more » « less
  2. ; ; ; ; ; ; ; ; (, Advanced Science)
    Abstract Gallium‐based liquid metal alloys (GaLMAs) have widespread applications ranging from soft electronics, energy devices, and catalysis. GaLMAs can be transformed into liquid metal emulsions (LMEs) to modify their rheology for facile patterning, processing, and material integration for GaLMA‐based device fabrication. One drawback of using LMEs is reduced electrical conductivity owing to the oxides that form on the surface of dispersed liquid metal droplets. LMEs thus need to be activated by coalescing liquid metal droplets into an electrically conductive network, which usually involves techniques that subject the LME to harsh conditions. This study presents a way to coalesce these droplets through a chemical reaction at mild temperatures (T∌ 80 °C). Chemical activation is enabled by adding halide compounds into the emulsion that chemically etch the oxide skin on the surface of dispersed droplets of eutectic gallium indium (eGaIn). LMEs synthesized with halide activators can achieve electrical conductivities close to bulk liquid metal (2.4 × 104S cm−1) after being heated. 3D printable chemically coalescing LME ink formulations are optimized by systematically exploring halide activator type and concentration, along with mixing conditions, while maximizing for electrical conductivity, shape retention, and compatibility with direct ink writing (DIW). The utility of this ink is demonstrated in a hybrid 3D printing process to create a battery‐integrated light emitting diode array, followed by a nondestructive low temperature heat activation that produces a functional device. 
    more » « less
  3. ; (, Journal of Computational and Graphical Statistics)
    We study the Bayesian multi-task variable selection problem, where the goal is to select activated variables for multiple related data sets simultaneously. We propose a new variational Bayes algorithm which generalizes and improves the recently developed “sum of single effects” model of Wang et al. (2020a). Motivated by differential gene network analysis in biology, we further extend our method to joint structure learning of multiple directed acyclic graphical models, a problem known to be computationally highly challenging. We propose a novel order MCMC sampler where our multi-task variable selection algorithm is used to quickly evaluate the posterior probability of each ordering. Both simulation studies and real gene expression data analysis are conducted to show the efficiency of our method. Finally, we also prove a posterior consistency result for multi-task variable selection, which provides a theoretical guarantee for the proposed algorithms. Supplementary materials for this article are available online. 
    more » « less
  4. ; ; (, Polymer Engineering & Science)
    ABSTRACT The processing and performance of multilayer films containing post‐consumer recycled polypropylene (rPP1 and rPP2) materials are investigated to understand the effect of layer multiplying elements (LMEs), die temperature, virgin polypropylene (vPP%) content, and polyethylene (PE) contamination in flexible packaging applications. Three‐layer coextruded films were created with virgin polypropylene (vPP) consistently applied as the outer layers while the core layer comprised recyclates with varying concentrations of polyethylene as an unintended contaminant to polypropylene. To enhance layer uniformity and interfacial interaction, a layer multiplying element (LME) was employed to increase the number of coextruded film layers from 3 to 9. Tensile properties (elongation at break, yield stress, and modulus) were characterized in both machine direction (MD) and transverse direction (TD); after which, multiple linear regression analyses were conducted on 45 observations to model the effect of each factor. The results indicated that the LME significantly enhanced elongation at break in TD by 1280% strain, while temperature and vPP fraction also contributed positively to ductility in TD (+341% and +2373%, respectively). However, PE contamination had a substantial negative impact on elongation in MD (−2449%) underscoring its embrittling role due to lack of compatibility with the PP matrix. Critically, LME partially mitigated the negative PE effect via an interaction term (PE*LME), improving elongation in MD by +3101%. Scanning Electron Microscopy (SEM) revealed a distinct, regular pattern of alternating polyethylene (PE) and polypropylene (PP) domains forming ribbon‐like fibrillar structures. This unique morphological arrangement suggests a self‐organizing behavior driven by immiscibility and flow‐induced alignment under extrusion conditions. The presence of regular alternating domains at near equal concentrations implies a balance among shear‐driven orientation, phase separation kinetics, and crystallization phenomena, resulting in an ordered micro‐fibrillar structure. Importantly, both monolayer and multilayer films containing rPP2 or rPP1/rPP2 blends exhibited these aligned, ribbon‐like fibrils oriented in the machine direction (MD). SEM analysis of fractured specimens further indicated that brittle failure was often associated with interfacial delamination, particularly in recyclate‐rich regions, whereas ductile failure exhibited entangled reinforcing fibrils, suggesting improved energy absorption and interlayer cohesion. Understanding and controlling this self‐organized microstructure could significantly enhance processing stability, mechanical properties, and potential applications of recycled polyolefin blends, offering novel strategies for tailoring recyclate morphology and performance. 
    more » « less
  5. (, Philosophical Transactions of the Royal Society B: Biological Sciences)
    Evolutionarily stable strategy (ESS) analysis pioneered by Maynard Smith and Price took off in part because it often does not require explicit assumptions about the genetics and demography of a population in contrast to population genetic models. Though this simplicity is useful, it obscures the degree to which ESS analysis applies to populations with more realistic genetics and demography: for example, how does ESS analysis handle complexities such as kin selection, group selection and variable environments when phenotypes are affected by multiple genes? In this paper, I review the history of the ESS concept and show how early uncertainty about the method lead to important mathematical theory linking ESS analysis to general population genetic models. I use this theory to emphasize the link between ESS analysis and the concept of invasion fitness . I give examples of how invasion fitness can measure kin selection, group selection and the evolution of linked modifier genes in response to variable environments. The ESSs in these examples depend crucially on demographic and genetic parameters, which highlights how ESS analysis will continue to be an important tool in understanding evolutionary patterns as new models address the increasing abundance of genetic and long-term demographic data in natural populations. This article is part of the theme issue ‘Half a century of evolutionary games: a synthesis of theory, application and future directions’. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email