Journal of Data Science

Accurate forecasting of student yield is critical for academic institutions because enrollment projections directly affect resource allocation, course offerings, housing capacity, and budget decisions. Existing enrollment models primarily focus on predicting individual matriculation probabilities, whereas institutional planning depends on accurate prediction of the aggregate enrollment count together with its uncertainty quantification. Here we develop statistical methods for aggregate enrollment prediction and interval estimation under logistic, regularized regression, tree-based, and ensemble classification models. For unpenalized logistic regression, we derive an asymptotic prediction interval and a parametric-simulation interval that propagates coefficient uncertainty. For regularized and tree-based learners, we develop a model-agnostic bootstrap interval that captures selection, estimation, and predictive uncertainty in a unified framework. To interpret predictive structure, we introduce partial and marginal AUC measures to quantify the unique and standalone contributions of feature groups. Simulation studies show that the proposed intervals attain close-to-nominal coverage under well-calibrated and bootstrap-stable learners. Applied to University of Connecticut in-state freshman cohorts from 2020–2025, the proposed intervals contain the observed 2025 enrollment count while identifying the feature groups that contribute most strongly to the enrollment prediction.

In 2022 the American Statistical Association established the Riffenburgh Award, which recognizes exceptional innovation in extending statistical methods across diverse fields. Simultaneously, the Department of Statistics at the University of Connecticut proudly commemorated six decades of excellence, having evolved into a preeminent hub for academic, industrial, and governmental statistical grooming. To honor this legacy, a captivating virtual dialogue was conducted with the department’s visionary founder, Dr. Robert H. Riffenburgh, delving into his extraordinary career trajectory, profound insights into the statistical vocation, and heartfelt accounts from the faculty and students he personally nurtured. This multifaceted narrative documents the conversation with more detailed background information on each topic covered by the interview than what is presented in the video recording on YouTube.

Preferential attachment (PA) network models have a wide range of applications in various scientific disciplines. Efficient generation of large-scale PA networks helps uncover their structural properties and facilitate the development of associated analytical methodologies. Existing software packages only provide limited functions for this purpose with restricted configurations and efficiency. We present a generic, user-friendly implementation of weighted, directed PA network generation with R package wdnet. The core algorithm is based on an efficient binary tree approach. The package further allows adding multiple edges at a time, heterogeneous reciprocal edges, and user-specified preference functions. The engine under the hood is implemented in C++. Usages of the package are illustrated with detailed explanation. A benchmark study shows that wdnet is efficient for generating general PA networks not available in other packages. In restricted settings that can be handled by existing packages, wdnet provides comparable efficiency.

Splines are important tools for the flexible modeling of curves and surfaces in regression analyses. Functions for constructing spline basis functions are available in R through the base package splines. When the curves to be modeled have known characteristics in monotonicity or curvature, more efficient statistical inferences are possible with shape-restricted splines. Such splines, however, are not available in the R package splines. The package splines2 provides easy-to-use shape-restricted spline basis functions, along with their derivatives and integrals which are important tools in many inference scenarios. It also provides additional splines and features that are not available in the splines package, such as periodic splines and generalized Bernstein polynomials. The usages of the functions are illustrated with shape-restricted regression, recurrent event data analysis, and extreme-value copulas.