About
Nicholas Clark is an Associate Professor in the Department of Mathematics at the University of St. Thomas in Saint Paul, Minnesota. His research centers on spatio-temporal statistical modeling, with current work on identifiability in spatio-temporal models for point process data.
He is a Co-PI on the NSF-funded SCORE with Data project, where he also serves as an editor for online module submissions. SCORE brings academics and sports-industry professionals together to build case-based learning materials that use sports data in undergraduate data science courses.
Before joining St. Thomas, he spent six years at the United States Military Academy at West Point, where he helped establish and later led the Applied Statistics and Data Science Program. While there, he helped create the U.S. Army's Data Literacy 101 program and developed approaches for assessing data literacy education across large organizations.
He has more than eight years of undergraduate teaching experience across more than 25 sections of seven courses, and is a trained ABET program evaluator for data science programs.
- Spatio-temporal point processes
- Statistics for defense and national security
- Data science program development and assessment
Selected recent publications
-
Abstract
Socioeconomically disadvantaged populations are often disproportionately subjected to over policing. While sometimes well intended, other times over policing is due to a belief that a response to crime must be swift in order to deter possible repeat actors. However, crime inspired by the action of another criminal is not the only reason why these events may be clustered in space and time. A competing theory states that spatio-temporal clustering occurs due to underlying socio-economic conditions rather than inspired actors. In quantitative criminology, repeat victimization attributed to copy-cat actors is often modeled through the use of a self-exciting, or Hawkes, process. This process is often assumed to exist prior to data analysis and alternative processes are rarely considered. In this manuscript, we will discuss how model selection, in particular model selection between a log Gaussian Cox process and a Hawkes process, is both a necessary as well as difficult step in statistical modeling of crime. We will provide a few techniques to conduct model selection between these processes and conclude, with a warning for researchers in this area, that sometimes these processes cannot be disentangled. In these instances, we suggest that modelers explicitly mention that their models rely on one theory of repeat victimization and that alternative theories may exist that lead to other forms of policing and may impact their interpretation of the root cause of why crime is spreading in space and time.
-
Abstract
In recent years, there has been an explosion in the growth of undergraduate statistics and data science programs across the US. Simultaneously, there has been clear guidance written on curriculum development for both data science (De Veaux et al.) and statistics (Carver et al.) programs. While this was occurring, ABET (now simply an acronym, but previously standing for the Accreditation Board for Engineering and Technology), in coordination with organizations such as the American Statistical Association, developed accreditation criteria for Data Science programs. In this article, we discuss our journey through ABET accreditation and discuss how adopting ABET processes for continuous improvement strengthens a program’s assessment process. We share best practices for working across multiple departments to collect data not only on individual courses, but also on the program as a whole. While the framework presented was initially established to support ABET accreditation, we argue that a properly executed program assessment should occur regardless of whether or not an institution is seeking ABET accreditation for their data science program. Throughout this article, we also discuss the extent to which ABET requirements naturally fit within our program’s existing goals, including an assessment of how ABET requirements align with major ideas in the field of data science education.
-
Abstract
Self-exciting models are statistical models of count data where the probability of an event occurring is influenced by the history of the process. In particular, self-exciting spatio-temporal models allow for spatial dependence as well as temporal self-excitation. For large spatial or temporal regions, however, the model leads to an intractable likelihood. An increasingly common method for dealing with large spatio-temporal models is by using Laplace approximations (LA). This method is convenient as it can easily be applied and is quickly implemented. However, as we will demonstrate in this manuscript, when applied to self-exciting Poisson spatial–temporal models, Laplace Approximations result in a significant bias in estimating some parameters. Due to this bias, we propose using up to sixth-order corrections to the LA for fitting these models. We will demonstrate how to do this in a Bayesian setting for self-exciting spatio-temporal models. We will further show there is a limited parameter space where the extended LA method still has bias. In these uncommon instances we will demonstrate how a more computationally intensive fully Bayesian approach using the Stan software program is possible in those rare instances. The performance of the extended LA method is illustrated with both simulation and real-world data.
-
Machine learning modeling practices to support the principles of AI and ethics in nutrition research
Abstract
Nutrition research is relying more on artificial intelligence and machine learning models to understand, diagnose, predict, and explain data. While artificial intelligence and machine learning models provide powerful modeling tools, failure to use careful and well-thought-out modeling processes can lead to misleading conclusions and concerns surrounding ethics and bias.
A complete list of publications is in my CV.