The North Carolina Directed Reading (NCDRP) program is a new extracurricular offering
launched by the NCSU Statistics Department in the Spring of 2027. The NCDRP is modeled
after other directed reading programs that have found success in other departments and
universities around the country.

This semester-long program is designed to help undergraduates build the skill of reading
academic papers while introducing them to exciting statistical concepts that go beyond the
classroom. Each NCDRP project will allow undergraduate participants to work with an NCSU
Statistics PhD student mentor and an NCSU Statistics faculty member through a course of
reading designed by the mentorship pair. The semester will begin with a kickoff event at which
the accepted students will meet their graduate student mentors, and will culminate in an
NCDRP end-of-semester event at which the undergraduate students will present on their
respective papers and areas of study.

Lastly, in launching this program we hope to contribute to a stronger statistical community at
NCSU by creating direct interpersonal connections between undergraduates, graduate students
and NCSU statistics faculty.

For any questions or concerns please contact: Owen Jacobs at [email protected]

We strongly encourage all interested students to apply here.

Student Research Projects

1. Cracks in the Foundations of Statistics

Student Name: Joseph Harrison

Faculty Advisor: Justin Post

Project Description: The undergraduate student will be exposed to foundational issues in statistical inference. The general learning objective is to grasp and articulate challenges which current statistical methodologies encounter, and how relaxations prove useful in tackling the statistical problem of inference.

Prerequisites: MA 141, ST 120, ST 121

Programming and Statistical Computing Prerequisites: None

Additional Qualifications (Preferred but Not Required): Strictly speaking, this paper should be approachable without ST422, although it will be challenging.

2. An Introduction to Stochastic Processes and Spatial Statistics

Student Name: Owen Jacobs

Faculty Advisor: Erin Schliep

Project Description: This project will introduce an undergraduate student to stochastic processes through the important role they play in spatial and environmental statistics. This will involve exposure to spatial statistics, hierarchical Bayesian modeling and a few examples of stochastic processes. The main focus will be Gaussian processes and their use in spatial statistics, motivated by environmental applications. The paper we will focus on covers a wide range of uses for Gaussian processes in spatial statistics and we can select which aspects to focus on based on the undergraduate student’s interests. For an undergraduate student interested in spatial statistics and the environment, this is a great place to start!

Prerequisites: MA 141, MA 241, ST 120, ST 121, ST 220

Programming and Statistical Computing Prerequisites: None

Additional Qualifications (Preferred but Not Required): It would be helpful (but not necessary) to have completed ST421, MA242 and have some past experience with a programming language like R or other statistical computing experience.

3. Counting the Invisible: Using Social Networks to Estimate Hidden Populations

Student Name: Colin Yip

Faculty Advisor: Nathaniel Josephs

Project Description: How do you estimate the number of people experiencing homelessness, living with a rare disease, or belonging to a hidden population when a traditional survey may never reach them? One surprising approach is to use the social networks that connect us. In this reading course, you will explore the Network Scale-Up Method, a family of statistical techniques that estimates the size of hard-to-reach populations by asking people about the individuals they know. You will learn how these estimators work, investigate the assumptions behind them, and discover how the structure of social networks can affect their accuracy. Along the way, you will be introduced to statistical network analysis and random graph models, gaining insight into how mathematical and statistical tools can help answer important real-world questions.

Prerequisites: MA 141, MA 241, ST 120, ST 121

Programming and Statistical Computing Prerequisites: Intro to a Data Science Toolkit with R and GitHub – ST 115

Additional Qualifications (Preferred but Not Required): None

4. Understanding Uncertainty in Machine Learning: An Introduction to Conformal Prediction

Student Name: Anqi Zhao

Faculty Advisor: Shu Yang

Project Description: Modern AI models and machine learning methods have achieved remarkable success across many areas of science, yet their underlying structures often remain black boxes. Conformal prediction (CP) provides a wrapper around essentially any machine learning model, offering model-agnostic prediction sets with finite-sample coverage guarantees. By joining this reading group, the student will read papers and learn the fundamentals of CP for both regression and classification. We will begin with a thorough introduction to basic CP methods and then focus more specifically on split conformal prediction. We will also discuss several extensions and applications of CP in different areas, such as LLM validity and counterfactual inference. By the end of the semester, the student will be expected to give a presentation on the theoretical foundations of CP and may also include one practical application of a CP method, either from one of the papers discussed in the reading group or from another paper they find. Depending on the student’s coding experience, they may also include simulations in their presentation, although this is not required.

Prerequisites: MA 305, ST 121, ST 220

Programming and Statistical Computing Prerequisites: No requirements on coding, but highly recommend either ST115 or ST216.

Additional Qualifications (Preferred but Not Required): Coding experiences in R or Python are highly recommended, equivalent to ST115 or ST216. Also prefer student with basic probability and regression, e.g., ST 371/372 or equivalent.

5. Exploring Model Performance Using a ‘Virtual Ecologist”

Student Name: Adira Cohen

Faculty Advisor: Erin Schliep

Project Description: One of the most fundamental concerns in statistics is the ability to select an appropriate model and understand what this model can and cannot do. Statisticians commonly rely on simulation studies to compare performance between models and across scenarios. This project will focus on ecological applications, which are by nature highly complex, forcing researchers to make strategic modeling choices based on their research goals. The student involved in this project will simulate an ecological system based on mechanistic processes from which a “virtual ecologist” will collect data; statistical models can then be fit using this data and their performance compared. For example, the student could create an agent-based model of a landscape through which deer move and are captured by camera traps at specified locations, and use these captures to fit various species distribution models. Deep understanding of the physical system and statistical models is necessary to design a simulation study which can identify strengths and weaknesses of these models.

The student involved in this project will gain experience in the following:

1. Formalize a research question and form an approach to address it, including designing a simulated ecological system with a “virtual ecologist” and selecting relevant statistical models.
2. Implement the chosen simulation study; this will likely involve creating an agent-based model and using existing packages to run statistical models.
3. Assess the performance of the chosen methods and think critically about why we observe these results.
4. Communicate recommendations for model choice and interpretation given a practitioner’s research goals.

The specific ecological application will be chosen based on the student’s interests, which will guide further readings. Many examples can be found in Table 1 of the main reading.

Prerequisites: ST 220, ST 240

Programming and Statistical Computing Prerequisites: Knowledge or willingness to learn a programming language which can be used to simulate an ecological system (e.g., NetLogo, R, Python) and run statistical models (e.g., R, Python, SAS).

Additional Qualifications (Preferred but Not Required): None