Search
You are now searching on the website "Department of Statistics". Would you like to search the entire Stockholm University website?
Search on su.seShowing 41-60 of 110 hits
- Seminar: Mattias Villani, Department of Statistics, Stockholm University
Department of Statistics
Bayesian modeling of effective and functional brain connectivity - an exercise in probabilistic programming Seminar Date: Wednesday 27 March 2024 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 28, house 4, level 2 Abstract A hierarchical vector autoregressive model is presented for analyzing differences in brain connectivity between autism spectrum disorder patients and healthy controls. This application will be used to demonstrate how probabilistic programming languages like Stan and Turing can solve quite complex applied statistical problems with state-of-the-art computational inference methodology. A brief introduction to variational inference for approximate large-scale Bayesian inference will also be given.
- Seminar: Benoit Liquet, School of Mathematical and Physical Sciences, Macquarie University, Sydney
Department of Statistics
Best Subset Selection via Continuous Optimization Seminar Date: Wednesday 12 June 2024 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 27, house 4, level 2 Abstract Recent rapid developments in information technology have enabled the collection of high-dimensional complex data, including in engineering, economics, finance, biology, and health sciences. High-dimensional means that the number of features is large and often far larger than the number of collected data samples (units). In many of these applications, it is desirable to find a small best subset of predictors so that the resulting model has desirable prediction accuracy. In this talk, we present the COMBSS framework, a continuous optimization-based solution that we recently showed to solve the best subset selection problem in linear regression. Then, we highlight how COMBSS can be extended to other models such as the logistic model. Finally, we present how to cast the best subset solution method into principal component analysis and partial least square frameworks.
- Thesis defence - Erik Spånberg
Department of Statistics
Ingress
- Seminar: Rolf Larsson, Uppsala University
Department of Statistics
Confidence distributions for the autoregressive parameter Seminar Date: Wednesday 11 October 2023 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 27, house 4, level 2 Abstract The notion of confidence distributions is applied to inference about the parameter in a simple autoregressive model, allowing the parameter to take the value one. This makes it possible to compare to asymptotic approximations in both the stationary and the non stationary cases at the same time. The main point, however, is to compare to a Bayesian analysis of the same problem. A non-informative prior for a parameter, in the sense of Jeffreys, is given as the ratio of the confidence density and the likelihood. In this way, the similarity between the confidence and non-informative Bayesian frameworks is exploited. It is shown that, in the stationary case, asymptotically the so induced prior is flat. However, if a unit parameter is allowed, the induced prior has to have a spike at one of some size. Simulation studies and two empirical examples illustrate the ideas.
- Seminar: Ingeborg Waernbaum, Uppsala University
Department of Statistics
Conditional independence criteria and sufficient selection sets for total and subpopulation treatment effects in the potential outcome framework Seminar Date: Wednesday 15 November 2023 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 27, house 4, level 2 Abstract When the study objective includes estimation of causal parameters, applying inclusion/exclusion criteria or the presence of missing data can bias the resulting estimator. In an observational study, a sufficient adjustment set for estimating a causal effect is a set such that confounding is adjusted for. Corresponding properties of covariate sets targeting selection bias have been shown in related work, within a structural casual framework with graphs. In this paper, this problem is revisited within the potential outcome framework. Here, we present sufficient conditions for identifying average causal effects in randomized and observational studies and propose data driven procedures for covariate adjustment safeguarding against selection bias. Joint work with Stina Zetterström.
- Seminar: Chamika Porage, Dept. of Statistics, Uppsala University
Department of Statistics
Prognostic score methods for the estimation of the average causal effects Seminar Date: Wednesday 20 November 2024 Time: 13.00 – 14.00 Location: Campus Albano, lecture room 28, house 4, level 2 Abstract The prognostic score (PGS) is a function of observed covariates that summarizes covariates’ association with potential responses. In the current study, we propose a full prognostic score (FPGS), an extension of the PGS that integrates individual prognostic scores to account for confounding adjustments in causal inference. Under effect modification, we demonstrate that FPGS meets the sufficiency condition for confounding adjustment. Consequently, the implemented FPGS is sufficient for estimating the average causal effect. To estimate PGS and FPGS, we apply linear regression, random forest regression, gradient boosting regression, and support vector machine. When estimating the average treatment effect, we incorporate FPGS into semi-parametric estimators, including regression imputation and targeted maximum likelihood estimation (TMLE). The finite sample properties of estimators are compared through three simulation studies. Based on the findings from FPGS estimators, the mean squared errors of the linear regression imputation estimator and TMLE estimator, which comprise linearly regressed PGS, are smaller than those of alternative estimators. In an empirical study, we analyze data from the National Health and Nutrition Examination Survey (NHANES, 2007-2008) to determine the effect of smoking on blood lead levels.
- Seminar: Nurgul Batyrbekova, Karolinska Institutet
Department of Statistics
Modelling multiple time-scales with flexible parametric survival models Seminar Date: Wednesday 29 January 2025 Time: 13.00 – 14.00 Location: Campus Albano, lecture room 25, house 4, level 2 Abstract There are situations when we need to model data with multiple time-scales. We usually use Cox or Poisson models to model data with multiple time-scales. This would require fine-splitting of the data to ensure that the assumption of proportional hazards within given time-intervals holds. However, this is not optimal when the dataset is large and when the models become too complex to manage. Iacobelli and Carstensen (2013) have shown how to fit multiple time-scales using Poisson models within a multistate framework. The drawback with Poisson method, is that users are required to assume constant hazard rates within time-intervals. An alternative approach is to fit flexible parametric survival models when modelling data with multiple time-scales. This method does not require assumtion of constant hazard rates, splitting data into small time-intervals, and therefore saves time, helps avoid technological limitations and reduces room for error.
- Seminar: Zangin Zeebari, Jönköpings University
Department of Statistics
Conditional Quantile Estimators of Unit Inefficiency in Stochastic Frontier Analysis Seminar Date: Wednesday 9 April 2025 Time: 13.00 – 14.00 Location: Campus Albano, lecture room 25, house 4, level 2 Abstract Having the ability to disentangle stochastic noise and firm/unit level inefficiency makes stochastic frontier analysis (SFA) an appealing technique for regulators. However, even with this ability, it is well known that commonly used conditional mean/mode estimators of unit inefficiency shrink towards the mean/mode of inefficiency rather than towards the actual level of unit inefficiency, especially in the presence of noise variance inflation. We propose conditional quantile estimators of unit inefficiency as an alternative and show analytically that they also shrink towards the quantiles of inefficiency rather than towards the level of unit inefficiency itself. However, we also show how advantageous this shrinkage property can be when we accurately estimate the quantile of each unit inefficiency and use its counterpart conditional quantile for estimating unit inefficiency. This allows for some degree of accuracy to be imparted to the analysis that is not available with the conditional mean/mode estimators of unit inefficiency due to their shrinkage property.
- Seminar: Stina Zetterström, University of Cambridge, UK
Department of Statistics
Alternative Formulations of the Allocation Probability Test Statistic to Enhance Power in Response-Adaptive Clinical Trials Seminar Date: Wednesday 8 October 2025 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 2, house 1, level 2 Abstract In response-adaptive clinical trials, treatment allocation probabilities are updated throughout the study based on the outcome of previous participants in the study. One reason for this strategy is to increase the probability that participants receive the most effective treatment compared to equal randomisation. However, imbalances in allocation can reduce statistical power when testing for treatment differences. Recent research has introduced a testing approach based on allocation probabilities (AP) rather than direct outcomes, which can improve power. I will present alternative AP test formulations to further increase power in response-adaptive settings. Using simulation studies with the Bayesian Response Adaptive Randomization (BRAR) algorithm for binary outcomes in a two-arm trial, the performance of the investigated functional forms is assessed. The results indicate that changing the functional form of the AP test statistic can substantially increase the power compared to the original formulation. Furthermore, the AP test can offer greater power than traditional methods while maintaining type I error control. Although the evaluation focuses on BRAR, the AP framework is applicable to any response-adaptive randomisation, and its performance can be investigated in those contexts.
- Seminar: Professor Yoshihisa Kashima, University of Melbourne
Department of Statistics
Cultivating a Culture of Sustainability - A Psychological Perspective Seminar Date: Friday 23 May 2025 Time: 14.00 – 15.15 Location: Campus Albano, Theatre 6, house 4, level 3
- Seminar: Måns Magnusson, Dept. of Statistics, Uppsala University
Department of Statistics
posteriordb: Testing, Benchmarking and Developing Bayesian Inference Algorithms Seminar Date: Wednesday 4 June 2025 Time: 13.00 – 14.00 Location: Campus Albano, lecture room 25, house 4, level 2 Abstract The general applicability and robustness of posterior inference algorithms is critical to widely used probabilistic programming languages such as Stan, PyMC, Pyro, and Turing.jl. When designing a new inference algorithm, whether it involves Monte Carlo sampling or variational approximation, the fundamental problem is evaluating its accuracy and efficiency across a range of representative target posteriors. To solve this problem, we propose posteriordb, a database of models and data sets defining target densities along with reference Monte Carlo draws. We further provide a guide to the best practices in using posteriordb for algorithm evaluation and comparison. To provide a wide range of realistic posteriors, posteriordb currently comprises 120 representative models with data, and has been instrumental in developing several inference algorithms.
- Seminar: Enoch Yi-Tung Chen, Karolinska Institutet
Department of Statistics
A Multistate Model Incorporating Relative Survival Extrapolation and Mixed Time Scales for Health Technology Assessment Seminar Date: Wednesday 22 October 2025 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 2, house 1, level 2 Abstract Multistate models have been widely applied in health technology assessment. However, extrapolating survival in a multistate model setting presents challenges in terms of accurate predictions. We develop an individual-level continuous time multistate model that integrates relative survival extrapolation and mixed time scales. We illustrate our proposed model using an illness-death model. We model the transition rates using flexible parametric models. We provide computer code to simulate event times from models with mixed time scales. This feature allows us to incorporate relative survival extrapolation in a multistate setting. We compare several multistate models with different parametric models (standard vs. flexible parametric models), and survival frameworks (all-cause vs. relative survival framework) using a previous clinical trial as an illustrative example. In the example case study, the results agreed better with the observed data than the commonly applied approach using standard parametric models within an all-cause survival framework. The model provides an alternative to combine short-term trial data with long-term external data within a multistate model context in health technology assessment. Reference: Chen EYT, Dickman PW, Clements MS. A multistate model incorporating relative survival extrapolation and mixed time scales for health technology assessment. PharmacoEconomics. 2025; 43:297–310. https://doi.org/10.1007/s40273-024-01457-w
- Seminar: Pär Stockhammar, Dept. of Statistics, Stockholm University
Department of Statistics
Econometrics in Swedish policymaking - Applications and reflections from 15 years of practice Seminar Date: Wednesday 5 November 2025 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 27, house 4, level 2 Abstract I will talk about how econometrics shapes policy decisions at major Swedish institutions, including the Ministry of Finance, National Institute of Economic Research and the Riksbank. For instance, I'll describe how they make their macroeconomic forecasts (which heavily affect policy decisions) and how econometrics has been used to answer policy-relevant questions. The presentation will be very non-technical and hopefully easily accessible also for non-experts.
- Seminar: Maximillian Mossfeldt, Redpine Stockholm
Department of Statistics
Teaching in the AI era - possibilities and challenges Seminar Date: Wednesday 19 November 2025 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 26, house 4, level 2. Abstract As AI models rapidly advance in capability, teachers face critical decisions about assessment and learning. What are current AI strengths and limitations, and what can we expect from future AI models? How should we adapt? Should we embrace AI tools for tasks such as creating and grading exams and theses, and if so, how to best do it? Are written exams the only way to ensure students aren't using AI tools, or are there alternatives? And how can we detect AI use in student writing and coding? Drawing on his AI expertise, Maximilian will shed light on these questions, offering insights rather than definitive answers.
- Seminar: Jonathan Januar, Melbourne School of Psychological Sciences, University of Melbourne
Department of Statistics
Models for missing data in covert networks Seminar Date: Thursday 30 May 2024 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 30, house 4, level 2 Abstract Missing data are a prominent issue wherever data can be collected and is an especially notable problem when dealing with network data. As network data describe dependencies between nodes, missing data can lead to a multitude of issues ranging from inaccurate estimates to differing inferential claims. When considering the secretive aspect of covert networks - such as criminal or terrorist networks - validated and reliable data are very difficult to obtain and thus leads to great concerns with missing and unreliable data in the discipline. This project aims to discuss the considerations and nuances for addressing missing data in covert networks. Despite missing data being discussed in the social network analysis literature, there is not enough detail or discussion to explicitly describe the patterns by which missing data are missing. We explore various empirical covert network biases and represent them using explicit models for mechanisms by which missing data are generated. We further extend models for missingness to models for the sampling of edges in a graph in order to represent empirical biases in the process by which covert networks are observed. Simulations of these models are applied to empirical covert network datasets to evaluate the effects of specifying explicit patterns of missingness. Ultimately, this project aims to bridge disconnects between statistical assumptions involved in addressing missingness in network analysis and the harsh empirical data collection conditions of covert networks by introducing statistically principled methods of generating plausible missingness mechanisms.
- Seminar: Per Gösta Andersson, Department of Statistics, Stockholm University
Department of Statistics
Confidence interval construction for a binomial p Seminar Date: Wednesday 22 May 2024 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 14, house 2, level 2 Abstract The Wald confidence interval has for a long time been the standard interval for a binomial p. Regrettably, it is still very much in use even if, as Phil Kott, RTI, so neatly has expressed it: “Everyone knows that the Wald interval stinks”. We will take a look at what other colleagues have to say and also at how our course books are dealing with this problem. Besides a display of the bad properties of the Wald interval, better alternatives will be presented. References Andersson, P.G. (2022). Approximate confidence intervals for a binomial p --- once again. Statistical Science 37, 598--606. Andersson, P.G. (2023). The Wald confidence interval for a binomial p as an illuminating ``bad" example. The American Statistician 77, 443--448. Andersson P.G. (2024). Confidence Interval for a Binomial p: Recommend the Wilson interval to Students, Chance 37, 13—15. Brown, L.D., Cai, T.T. and DasGupta, A. (2001). Interval estimation for a binomial proportion. Statistical Science 16, 101--117. Brown, L.D., Cai, T.T. and DasGupta, A. (2002). Confidence intervals for a binomial proportion and asymptotic expansions. The Annals of Statistics 30, 160--201.
- Seminar: Gebrenegus Ghilagaber, Department of Statistics, Stockholm University
Department of Statistics
Bayesian Ordinal Modeling of Responses on the Likert-Scale: Illustration with Effects of Online Teaching on Students with Special Needs Seminar Date: Wednesday 22 November 2023 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 26, house 4, level 2 Abstract We demonstrate the power of Bayesian approach to modeling data with ordered outcome variable and illustrate it using data on the effect of the imposed online teaching during the Covid-19 pandemic on students with special needs. It is known that even frequentist ordinal modeling provides better theoretical interpretation and inference for such data than models which treat the ordered (Likert) outcome variable as metric. We argue that Bayesian approach adds flexibility in specifying the ordered models and provides goodness-of-fit measures to compare alternative models such as cumulative model and adjacent-category model with equal and unequal variances across levels of a covariate. Data for illustration come from about 6000 students who responded to a questionnaire on online teaching at Stockholm University during the Spring term 2021. Preliminary results using the Bayesian Regression Models using 'Stan’ (brms) R-Package show that students with special needs were adversely affected by the imposed online teaching. Among other things, the students felt less engaged intellectually, had lower confidence, felt examination instructions were unclear and spent more hours per week than their counterparts with no special needs.
- Seminar: Dan Hedlin, Department of Statistics, Stockholm University
Department of Statistics
A computed 95% confidence interval does cover the true value with probability 0.95 – epistemically interpreted Seminar Date: Wednesday 31 January 2024 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 9, house 2, level 2 Abstract How to interpret and make use of a computed (realised) confidence interval has been debated for a long time. I advocate that a realised confidence interval covers the parameter with probability . Suppose a sample of batteries in routine use is taken and their lifetimes are measured. A confidence interval is computed to days. The standard interpretation is that if we repeatedly draw samples of the same size and compute confidence intervals, 95% of the intervals will cover the unknown true life time. What can be said about the realised interval has not been clear in the literature. I shall give contradictory examples of interpretations. To appreciate the claim made here, we need to have a look at philosophy of probability and the Fisherian concept of relevant subsets.
- Seminar: Martin Hyllienmark, Department of Statistics, Stockholm University
Department of Statistics
Selection bias in big data in official statistics from a practitioner’s point of view Seminar Date: Wednesday 28 February 2024 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 15, house 2, level 2 Abstract Which is better: simple random sampling (SRS) with nonresponse or a large dataset with selection bias? Selection bias is increasingly problematic in surveys. Inspired by Meng’s (2018) paper Statistical paradises and paradoxes in big data, where he highlights statistical issues that bigness of data sets incur, we simulated sequences of growing populations with two different sampling designs: simple random sampling (SRS) with nonresponse and nonprobability data with selection bias (“big data”). The results showed that nonresponse combined with selection bias led to a constant bias and MSE only if the proportion of respondents was constant as the population size grew. Furthermore, there was a trade-off between bias and coverage probability caused by the amount of data available. By applying weights, it was possible to improve the results. The results in this study suggest that the data sources of choice may depend on the application of the statistics. Tools for comparing different data sources were investigated and discussed from a practical point of view.
- Seminar: Katarzyna Filipiak, Institute of Mathematics, Poznan University of Technology
Department of Statistics
Projection as the maximum likelihood estimator of the covariance structure Seminar Date: Wednesday 6 March 2024 Time: 13.00 – 14.00 Location: Campus Albano, Lecture room 27, house 4, level 2 Abstract Under the multivariate model with linearly structured covariance matrix with unknown variance components and known mean parameters it is known that the maximum likelihood estimators (MLEs) of variance components have explicit representation if and only if the space of covariance matrix is a quadratic subspace. In this talk, it will be shown, that the above result can be extended also for models with unknown mean parameters. Moreover, it will be proven that is such a model the projection of the MLE of unstructured covariance onto the space of structured matrices is still the MLEs as long as the estimated structure belongs to quadratic space.