(NMST611) Advanced Statistical Seminar -- Summer Term 2025
Wednesday: 15:40 - 17:20 | Prezenčne v Praktiku KPMS
The advanced statistical seminar consists of presentations delivered (typically in person) by invited foreign speakers or departmental guests. Assorted topics from modern statistics -- theory and applications -- are usually communicated during the talks.
Seminar schedule (Summer term 2025)
- 19.02.2025 | 15:40 | Robert BAJONS
WU Vienna University of Economics and Business, Austria
Title: Soccer analytics: A statistical view on xG and GAX
Expected Goals (xG) is one of the most popular metrics in modern football (soccer) analytics. xG models assign a probability of success to each shot, by relating it to shot-specific covariates. Popular xG models are typically based on high-level machine learning models that account for non-linear and interaction effects of the shot-specific covariates. As a measure of a shot’s value, it is commonly used to evaluate the shooting skills of players by considering goals over expectation (GAX), i.e. the difference between actual and expected goal for each shot. However, GAX is often criticized for being unstable over seasons and for not providing (direct) means of uncertainty quantification. In this work, we address both issues by showing how the player-specific GAX relates to a score test when the xG model is a logistic regression and proposing a natural nonparametric extension that enjoys doubly robustness properties and can be used with any sufficiently powerful machine learning algorithm for xG. In this way, we are able to leverage commonly used black-box xG models, while still obtaining valid statistical inferences on the player-specific odds or probability of scoring a goal, i.e. on the influence of a player on shooting the ball. Moreover, in order to make the results more accessible for practitioners, we show how the proposed procedure can yield player-specific effect estimates in a partially linear logistic regression model which are interpretable as additive effects on the log-odds of scoring a goal from a shot. Finally, we apply our framework to the 2015/16 season of the top five European leagues, determine the best shooters, and compare results across state-of-the-art xG models.
------------------------------------------------------------------------------------------- - 05.03.2025 | 15:40 | Lukas SOMMEREGGER
Alpen-Adria-Universität Klagenfurt, Austria
Title: The Road to Reliability: Lifetime Models for Automotive Semiconductors
The talk will present an overview of advanced reliability modeling techniques for automotive semiconductors, with a focus on stress test data analysis of discrete panel data, guard banding, and lifetime reliability prediction. We will discuss the application of non-parametric Markov chain estimations and extensions to tensor and non-linear conditional models to model complex system behavior. We further will explore the use of regularization techniques to improve the accuracy of continuous models when applied to discrete data. Additionally, we will present methods for optimizing guard bands to balance reliability and performance, and introduce an integrated formulation for predicting future unobserved values. Finally, we will also discuss the implications of these techniques for the health management and testing of reliable automotive semiconductors in the context of autonomous vehicles.
------------------------------------------------------------------------------------------- - 19.03.2025 | 15:40 | Gábor SZŰCS
Comenius University Bratislava, Slovakia
Title: Mathematical and statistical modeling of the volatility process of selected investment instrument returns
The talk will be devoted to mathematical and statistical modeling of stochastic volatility processes to better understand the dynamics of returns for selected investment instruments. The primary motivation for this research stems from the observation that fluctuations in investment returns rarely follow a normal probability distribution. Empirical studies have shown that their variations exhibit high kurtosis and are characterized by probability distributions with heavier tails than the normal distribution. Parameter estimations were obtained in the R software, using Markov Chain Monte Carlo (MCMC) and Hamiltonian Monte Carlo (HMC) techniques. The efficiency of the calibrated models was assessed using information criteria such as the deviance information criterion (DIC) and the so-called widely applicable information criterion (WAIC).
------------------------------------------------------------------------------------------- - 02.04.2025 | 15:40 | Ignacio CASCOS
Universidad Carlos III de Madrid, Spain
Title: Expectile depth: strictly positive variants, with applications
In multivariate statistics, depth refers to the degree of centrality of a point relative to a probability distribution or data cloud. Various notions of depth have been introduced in the statistical literature and applied in nonparametric data analysis techniques, such as supervised and unsupervised classification, statistical process control, and data imputation. For a univariate distribution, the expectile at level alpha is the solution to an asymmetric least squares minimization problem which is similar to the one that defines the alpha-quantile but instead relies on quadratic loss. The expectile depth of a point in a multivariate sample is defined as the minimum level, across all univariate projections, at which the projected point corresponds to the expectile at level alpha of the projected dataset. We will discuss theoretical and computational aspects of expectile depth, along with its applications in data visualization. A notable limitation of expectile depth — shared by most classical depth notions, such as halfspace, simplicial, and zonoid depth — is that it vanishes for points lying outside the convex hull of the data cloud. To address this issue, we introduce variants of expectile depth that remain strictly positive across the entire Euclidean space. Finally, we briefly explore how these constructions can be applied to the assessment of biological age.
------------------------------------------------------------------------------------------- - 16.04.2025 | 15:40 | Yvik SWAN
Université Libre de Bruxelles, Belgium
Title: Some Stein operators for multivariate distributions and applications
The title of this talk is basically its abstract : we present some Stein operators for multivariate distributions, along with their applications. The first application concerns the very classical problem of multivariate normal approximation; the second application concerns parameter estimation for Fisher-Bingham distributions on the sphere.
------------------------------------------------------------------------------------------- - 30.04.2025 | 15:40 | Mark PODOLSKIJ
University of Luxembourg, Luxembourg
Title: On nonparametric estimation of the interaction function in particle system models
This talk delves into the challenging problem of nonparametric estimation for the interaction function within diffusion-type particle system models. We introduce two estimation methods based on empirical risk minimization. Our study encompasses an analysis of the stochastic and approximation errors associated with both procedures, along with an examination of certain minimax lower bounds. In particular, for the first method we show that there is a natural metric under which the corresponding estimation error of the interaction function converges to zero with a parametric rate that is minimax optimal. This result is rather surprising given the complexity of the underlying estimation problem and a rather large class of interaction functions for which the above parametric rate holds.
------------------------------------------------------------------------------------------- - 14.05.2025 | 15:40 | Bei JIANG
University of Alberta, Canada
Title: TBA
...
-------------------------------------------------------------------------------------------
Advanced Statistical Seminar (Archiv)
|
The archive of the guests (invited speakers) of the Advanced statistical seminar (NMST611)
from previous semesters together with the title of the talks and short abstracts. |
|
| Winter term 2024/2025 | Summer term 2024 |
|---|---|
| Winter term 2023/2024 | Summer term 2023 |
| Summer term 2022 | |
| Summer term 2021 | |
