HUB
https://www.stat.washington.edu/taxonomy/term/486
enMultiplicative Models for Register Data
https://www.stat.washington.edu/event/seminar/multiplicative-models-register-data
<span>Multiplicative Models for Register Data</span>
<span><span lang="" about="https://www.stat.washington.edu/user/114" typeof="schema:Person" property="schema:name" datatype="" xml:lang="">kyunchan</span></span>
<span>Wed, 03/13/2019 - 09:08</span>
<div class="field field--name-field-speaker field--type-entity-reference field--label-above">
<div class="field--label">Speaker</div>
<div class="field--items">
<div class="field--item">Tamas Rudas</div>
</div>
</div>
<div class="field field--name-body field--type-text-with-summary field--label-hidden field--item"><p>Registers are increasingly important sources of data to be analyzed. Examples include registers of congenital abnormalities, supermarket purchases, or traffic violations. In such registers, records are created when a relevant event is observed, and they contain the features characterizing the event. Understanding the structure of associations among the features is of primary interest. However, the registers often do not contain cases in which no feature is present and therefore, standard multiplicative or log-linear models may not be applicable. The number of babies born without any anomaly is known, and the register can be completed. Supermarket visits without any purchase do occur, although their number is unknown. But traffic violations when no rule is violated make no sense, and models should not imply an estimate for their number. For cases, when quasi models do not provide a good fit, the talk presents hierarchical multiplicative models built on restricting the values of non-homogeneous odds ratios and describes their relationship with quasi log-linear models, using concepts of algebraic statistics. This is joint work with Anna Klimova.</p></div>
<div class="field field--name-field-summary-image field--type-image field--label-above">
<div class="field--label">Summary Image</div>
<div class="field--item"> <img src="https://www.stat.washington.edu/sites/default/files/2019-04/Rudas%2C%20Tamas-1_0.jpg" width="1000" height="1000" alt="" typeof="foaf:Image" class="img-responsive" /></div>
</div>
Wed, 13 Mar 2019 16:08:32 +0000kyunchan5446 at https://www.stat.washington.eduEssential Regression
https://www.stat.washington.edu/event/seminar/essential-regression
<span>Essential Regression</span>
<span><span lang="" about="https://www.stat.washington.edu/user/114" typeof="schema:Person" property="schema:name" datatype="" xml:lang="">kyunchan</span></span>
<span>Wed, 03/13/2019 - 08:19</span>
<div class="field field--name-field-speaker field--type-entity-reference field--label-above">
<div class="field--label">Speaker</div>
<div class="field--items">
<div class="field--item">Florentina Bunea</div>
</div>
</div>
<div class="field field--name-body field--type-text-with-summary field--label-hidden field--item"><p>We introduce the Essential Regression model, which provides an alternative to the ubiquitous <em>K</em>-sparse high dimensional linear regression on p variables. While K-sparse regression assumes that only <em>K</em> components of the observable <em>X</em> directly influence Y , Essential Regression allows for all components of <em>X </em>to influence <em>Y </em>, but mediated through a K-dimensional random vector <em>Z</em>. The mediator Z is unobservable, and made <em>interpretable </em>via a modeling assumption through which each component of <em>Z </em>is given the physical meaning of a small group of the <em>X</em>-variables. Essential Regression is the regression of <em>Y</em> on such <em>Z </em>∈ ℝᴷ. It is a high dimensional modeling tool, that can be viewed as a regression model at a coarser resolution level than that provided by regression on the more detailed, <em>p</em>-dimensional vector <em>X</em>. </p>
<p>Formally, E-Regression is a new type of latent factor regression model, in which the unobservable factors <em>Z </em>influence linearly both the response Y and the covariates <em>X</em>. Its novelty consists in the conditions that give <em>Z </em>interpretable meaning as well as render the regression coefficients <em>β </em>∈ ℝᴷ relating <em>Y </em>to <em>Z </em>— along with other important parameters of the model — identifiable. </p>
<p>The interpretability of the latent factors <em>Z</em> in Essential Regression allows us to provide conceptually new inferential tools for regression in high dimensions. In particular, <em>K</em>-dimensional inference at the Z<em> </em>level is a viable alternative to existing approaches that benefits from the greater simplicity and lower dimensionality of the “essence” <em>Z </em>when compared to <em>X</em>. It is furthermore well known that inference performed in <em>K</em>-sparse regression — after consistent support recovery and estimation of <em>K </em>— is valid only for the large regression coefficients of <em>Y </em>on <em>X</em>, which makes this approach problematic in practice. In contrast, inference performed in regression at the lower resolution level given by <em>Z </em>is uniform over the space of the regression coefficients <em>β </em>∈ ℝᴷ, although it is performed after estimating consistently <em>K </em>and the subset of the X-variables that explain <em>Z</em>. For this we construct computationally efficient estimators of <em>β</em>, derive their minimax rate, and show that they are minimax-rate optimal in Euclidean norm for every sample size <em>n</em>. We show that the component-wise estimates of <em>β </em>are asymptotically normal, with small asymptotic variance. This is a new addition to the literature in factor models, in which the inference problem is under-explored. </p>
<p>Prediction of <em>Y </em>from <em>X </em>under E-Regression complements, in the low signal to noise ratio regime, the battery of methods developed for prediction under other factor model specifications. Similarly to other methods, it is particularly powerful when <em>p </em>is large, with further refinements made possible by the Essential Regression model specifications. </p>
<p>E-Regression also provides a statistical framework for analysis in regression with clustered predictors, with or without overlap. This allows us to address possible inferential questions in post clustering-inference, and subsequently provide guidelines regarding the use and misuse of cluster averages as very popular dimension reduction devices in high dimensional regression.</p></div>
<div class="field field--name-field-summary-image field--type-image field--label-above">
<div class="field--label">Summary Image</div>
<div class="field--item"> <img src="https://www.stat.washington.edu/sites/default/files/2019-03/Bunea%2C%20Florentina.jpg" width="1000" height="1000" alt="" typeof="foaf:Image" class="img-responsive" /></div>
</div>
Wed, 13 Mar 2019 15:19:58 +0000kyunchan5441 at https://www.stat.washington.eduInstrumental Variable Learning of Marginal Structural Models
https://www.stat.washington.edu/event/seminar/instrumental-variable-learning-marginal-structural-models
<span>Instrumental Variable Learning of Marginal Structural Models</span>
<span><span lang="" about="https://www.stat.washington.edu/user/114" typeof="schema:Person" property="schema:name" datatype="" xml:lang="">kyunchan</span></span>
<span>Wed, 01/02/2019 - 11:59</span>
<div class="field field--name-field-speaker field--type-entity-reference field--label-above">
<div class="field--label">Speaker</div>
<div class="field--items">
<div class="field--item">Eric J. Tchetgen Tchetgen</div>
</div>
</div>
<div class="field field--name-body field--type-text-with-summary field--label-hidden field--item"><p>In a seminal paper, Robins (1998) introduced marginal structural models (MSMs), a general class of counterfactual models for the joint effects of time-varying treatment regimes in complex longitudinal studies subject to time-varying confounding. He established identification of MSM parameters under a sequential randomization assumption (SRA), which rules out unmeasured confounding of treatment assignment over time. We extend Robins' MSM theory by considering identification of MSM parameters with the aid of a time-varying instrumental variable, when sequential randomization fails to hold due to unmeasured confounding. Our identification conditions essentially require that no unobserved confounder predicts compliance type at each follow-up time. Under this assumption, we obtain a large class of semiparametric estimators that extends standard inverse-probability weighting (IPW) and includes multiply robust estimators, including a locally semiparametric efficient estimator. The approach provides a unified solution to IV inference from point exposure to time-varying exposure settings, including mean models with possibly nonlinear link functions, quantile MSMs and time to event models such as Cox MSMs. Finally, we briefly discuss recent robust IV methods that further allow for violation of the core IV identifying condition, the exclusion restriction assumption, without compromising inference.</p></div>
<div class="field field--name-field-summary-image field--type-image field--label-above">
<div class="field--label">Summary Image</div>
<div class="field--item"> <img src="https://www.stat.washington.edu/sites/default/files/2019-01/Tchetgen%20Tchetgen%2C%20Eric.jpg" width="1000" height="1000" alt="" typeof="foaf:Image" class="img-responsive" /></div>
</div>
Wed, 02 Jan 2019 19:59:13 +0000kyunchan5408 at https://www.stat.washington.edu[CANCELLED] Statistical Methods for Two Problems in Biology
https://www.stat.washington.edu/event/seminar/cancelled-statistical-methods-two-problems-biology
<span>[CANCELLED] Statistical Methods for Two Problems in Biology</span>
<span><span lang="" about="https://www.stat.washington.edu/user/114" typeof="schema:Person" property="schema:name" datatype="" xml:lang="">kyunchan</span></span>
<span>Wed, 01/02/2019 - 11:45</span>
<div class="field field--name-field-speaker field--type-entity-reference field--label-above">
<div class="field--label">Speaker</div>
<div class="field--items">
<div class="field--item">Daniela Witten</div>
</div>
</div>
<div class="field field--name-body field--type-text-with-summary field--label-hidden field--item"><p><em><strong>Note 2/7/2018: </strong></em><em><strong><span>We are canceling this seminar as a precaution in anticipation of the expected Winter storm.</span></strong></em></p>
<p> </p>
<p>As the pace and scale of data collection continues to increase across all areas of biology, there is a growing need for effective and principled statistical methods for the analysis of the resulting data. In this talk, I'll describe two ongoing projects to help fill this gap. </p>
<p>First, calcium imaging data is transforming the field of neuroscience by making it possible to assay the activities of large numbers of neurons simultaneously. For each neuron, the resulting "fluorescence trace" can be seen as a noisy surrogate of its spikes over time. In order to deconvolve a fluorescence trace into the underlying spike times, we consider an auto-regressive model for calcium dynamics. This leads naturally to a seemingly intractable $\ell_0$ optimization problem. I will show that it is in fact possible to efficiently solve this optimization problem for the global optimum, leading to substantial improvements over competing approaches.</p>
<p>Second, across many areas of biology, it is becoming increasingly common to collect "multi-view data": that is, data in which multiple data types (e.g. gene expression, DNA sequence, clinical measurements) have been measured on a single set of observations (e.g. patients). I will consider the following question: given a set of n observations with measurements on L data types, can a single clustering of the n observations be defined on all L data types, or does each data type have its own clustering of the observations? To answer this question, I will introduce a general framework for modeling multi-view data, as well as hypothesis tests that can be used in order to characterize the extent to which the clusterings on each of the L data types are the same or different.</p>
<p>This is joint work with PhD students Lucy Gao and Sean Jewell at UW, as well as Jacob Bien (University of Southern California), Paul Fearnhead (Lancaster), and Toby Hocking (Northern Arizona University). </p></div>
<div class="field field--name-field-summary-image field--type-image field--label-above">
<div class="field--label">Summary Image</div>
<div class="field--item"> <img src="https://www.stat.washington.edu/sites/default/files/2019-01/Witten%2C%20Daniela_0.jpg" width="1000" height="1000" alt="" typeof="foaf:Image" class="img-responsive" /></div>
</div>
Wed, 02 Jan 2019 19:45:06 +0000kyunchan5406 at https://www.stat.washington.eduA New Standard for the Analysis and Design of Replication Studies
https://www.stat.washington.edu/event/seminar/new-standard-analysis-and-design-replication-studies
<span>A New Standard for the Analysis and Design of Replication Studies</span>
<span><span lang="" about="https://www.stat.washington.edu/user/114" typeof="schema:Person" property="schema:name" datatype="" xml:lang="">kyunchan</span></span>
<span>Wed, 01/02/2019 - 11:18</span>
<div class="field field--name-field-speaker field--type-entity-reference field--label-above">
<div class="field--label">Speaker</div>
<div class="field--items">
<div class="field--item">Leonhard Held</div>
</div>
</div>
<div class="field field--name-body field--type-text-with-summary field--label-hidden field--item"><p>A new standard is proposed for the evidential assessment of replication studies. The approach combines a specific reverse-Bayes technique with prior-predictive tail probabilities to define replication success. The method gives rise to a quantitative measure for replication success, called the sceptical p-value. The sceptical p-value integrates traditional significance of both the original and replication study with a comparison of the respective effect sizes. It incorporates the uncertainty of both the original and replication effect estimates and reduces to the ordinary p value of the replication study if the uncertainty of the original effect estimate is ignored. The proposed framework can also be used to determine the power or the required sample size to achieve replication success. Numerical calculations highlight the difficulty to achieve replication success if the evidence from the original study is only suggestive. An application to data from the Open Science Collaboration project on the replicability of psychological science illustrates the proposed methodology.</p></div>
<div class="field field--name-field-summary-image field--type-image field--label-above">
<div class="field--label">Summary Image</div>
<div class="field--item"> <img src="https://www.stat.washington.edu/sites/default/files/2019-01/Held%2C%20Leonhard.jpg" width="1000" height="1000" alt="" typeof="foaf:Image" class="img-responsive" /></div>
</div>
Wed, 02 Jan 2019 19:18:21 +0000kyunchan5403 at https://www.stat.washington.eduHow Statistics Took Me to the Aleutian Islands
https://www.stat.washington.edu/event/seminar/how-statistics-took-me-aleutian-islands
<span>How Statistics Took Me to the Aleutian Islands</span>
<span><span lang="" about="https://www.stat.washington.edu/user/114" typeof="schema:Person" property="schema:name" datatype="" xml:lang="">kyunchan</span></span>
<span>Thu, 10/04/2018 - 15:36</span>
<div class="field field--name-field-speaker field--type-entity-reference field--label-above">
<div class="field--label">Speaker</div>
<div class="field--items">
<div class="field--item">Joel Howard Reynolds</div>
</div>
</div>
<div class="field field--name-body field--type-text-with-summary field--label-hidden field--item"><p>Did you know that your skills in statistics can be applied to ensure natural resources, such as fish, wildlife and even ecosystems, remain resilient into the future? That your love of algebra can take you to wild, remote, and amazing places? That there are careers where you get to collaborate with a wide variety of dedicated scientists working to better understand the world, how it is changing, and what it will be like in the future? A career as an applied statistician in a natural resource agency can be both intellectually stimulating and personally rewarding as you help inform and improve resource management for the benefit of current and future generations. This talk will touch on the importance of Statistical Thinking in resource management; the types of agency positions filled by applied statisticians and the roles they play; important knowledge and skills required to successfully hold such positions, and where to learn more about opportunities in this field. Examples will be drawn from the speaker’s ~20 years of experience in Alaska.</p></div>
<div class="field field--name-field-summary-image field--type-image field--label-above">
<div class="field--label">Summary Image</div>
<div class="field--item"> <img src="https://www.stat.washington.edu/sites/default/files/2018-10/Reynolds%2C%20Joel.jpg" width="1000" height="1000" alt="" typeof="foaf:Image" class="img-responsive" /></div>
</div>
Thu, 04 Oct 2018 22:36:03 +0000kyunchan5277 at https://www.stat.washington.eduBayesian Approaches to Dynamic Model Selection
https://www.stat.washington.edu/event/seminar/bayesian-approaches-dynamic-model-selection
<span>Bayesian Approaches to Dynamic Model Selection</span>
<span><span lang="" about="https://www.stat.washington.edu/user/114" typeof="schema:Person" property="schema:name" datatype="" xml:lang="">kyunchan</span></span>
<span>Tue, 09/25/2018 - 10:21</span>
<div class="field field--name-field-speaker field--type-entity-reference field--label-above">
<div class="field--label">Speaker</div>
<div class="field--items">
<div class="field--item">Michele Guindani</div>
</div>
</div>
<div class="field field--name-body field--type-text-with-summary field--label-hidden field--item"><p>In many applications, investigators monitor processes that vary in space and time, with the goal of identifying temporally persistent and spatially localized departures from a baseline or ``normal" behavior. In this talk, I will first discuss a principled Bayesian approach for estimating time varying functional connectivity networks from brain fMRI data. Dynamic functional connectivity, i.e., the study of how interactions among brain regions change dynamically over the course of an fMRI experiment, has recently received wide interest in the neuroimaging literature. Our method utilizes a hidden Markov model for classification of latent neurological states, achieving estimation of the connectivity networks in an integrated framework that borrows strength over the entire time course of the experiment. Furthermore, we assume that the graph structures, which define the connectivity states at each time point, are related within a super-graph, to encourage the selection of the same edges among related graphs. Then, I will propose a Bayesian nonparametric model selection approach with an application to the monitoring of pneumonia and influenza (P&I) mortality, to detect influenza outbreaks in the continental United States. More specifically, we introduce a zero-inflated conditionally identically distributed species sampling prior which allows borrowing information across time and to assign data to clusters associated to either a null or an alternate process. Spatial dependences are accounted for by means of a Markov random field prior, which allows to inform the selection based on inferences conducted at nearby locations. We show how the proposed modeling framework performs in an application to the P&I mortality data and in a simulation study, and compare with common threshold methods for detecting outbreaks over time, with more recent Markov switching based models, and with other Bayesian nonparametric priors that do not take into account spatio-temporal dependence.</p></div>
<div class="field field--name-field-summary-image field--type-image field--label-above">
<div class="field--label">Summary Image</div>
<div class="field--item"> <img src="https://www.stat.washington.edu/sites/default/files/2018-11/Guindani%2C%20Michele.jpg" width="1000" height="1000" alt="" typeof="foaf:Image" class="img-responsive" /></div>
</div>
Tue, 25 Sep 2018 17:21:01 +0000kyunchan5273 at https://www.stat.washington.eduSpectral Gap in Random Bipartite Biregular Graphs and Applications
https://www.stat.washington.edu/event/seminar/spectral-gap-random-bipartite-biregular-graphs-and-applications
<span>Spectral Gap in Random Bipartite Biregular Graphs and Applications</span>
<span><span lang="" about="https://www.stat.washington.edu/user/114" typeof="schema:Person" property="schema:name" datatype="" xml:lang="">kyunchan</span></span>
<span>Tue, 09/25/2018 - 10:18</span>
<div class="field field--name-field-speaker field--type-entity-reference field--label-above">
<div class="field--label">Speaker</div>
<div class="field--items">
<div class="field--item">Ioana Dumitriu</div>
</div>
</div>
<div class="field field--name-body field--type-text-with-summary field--label-hidden field--item"><p><span><span><span>The asymptotics of the second-largest eigenvalue in random regular graphs (also referred to as the "Alon conjecture") have been computed by Joel Friedman in his celebrated 2004 paper. Recently, a new proof of this result has been given by Charles Bordenave, using the non-backtracking operator and the Ihara-Bass formula. In the same spirit, we have been able to translate Bordenave's ideas to bipartite biregular graphs in order to calculate the asymptotical value of the second-largest pair of eigenvalues, and obtained a similar spectral gap result. Applications include community detection in equitable graphs or frames, matrix completion, and the construction of channels for efficient and tractable error-correcting codes (Tanner codes). This work is joint with Gerandy Brito and Kameron Harris.</span></span></span></p></div>
<div class="field field--name-field-summary-image field--type-image field--label-above">
<div class="field--label">Summary Image</div>
<div class="field--item"> <img src="https://www.stat.washington.edu/sites/default/files/2018-10/Dumitriu%2C%20Ioana_1.jpg" width="1000" height="1000" alt="" typeof="foaf:Image" class="img-responsive" /></div>
</div>
Tue, 25 Sep 2018 17:18:11 +0000kyunchan5269 at https://www.stat.washington.eduFast Inference for Spatial Generalized Linear Mixed Models
https://www.stat.washington.edu/event/seminar/fast-inference-spatial-generalized-linear-mixed-models
<span>Fast Inference for Spatial Generalized Linear Mixed Models</span>
<span><span lang="" about="https://www.stat.washington.edu/user/68" typeof="schema:Person" property="schema:name" datatype="" xml:lang="">trpham</span></span>
<span>Mon, 03/26/2018 - 09:09</span>
<div class="field field--name-field-speaker field--type-entity-reference field--label-above">
<div class="field--label">Speaker</div>
<div class="field--items">
<div class="field--item">Murali Haran</div>
</div>
</div>
<div class="field field--name-body field--type-text-with-summary field--label-hidden field--item"><p><span><span><span>Non-Gaussian spatial data arise in a number of disciplines. Examples include spatial data on disease incidences (counts), and satellite images of ice sheets (presence-absence). Spatial generalized linear mixed models (SGLMMs), which build on latent Gaussian processes or Markov random fields, are convenient and flexible models for such data and are used widely in mainstream statistics and other disciplines. For high-dimensional data, SGLMMs present significant computational challenges due to the large number of dependent spatial random effects. I will discuss projection-based approaches that reparameterize and reduce the number of random effects in SGLMMs, resulting in a dramatic reduction in computational costs for Bayesian and maximum likelihood inference. Our approach also addresses spatial confounding issues. This talk is based on joint work with Yawen Guan (SAMSI) and John Hughes (U of Colorado-Denver).</span></span></span></p></div>
Mon, 26 Mar 2018 16:09:19 +0000trpham5075 at https://www.stat.washington.eduStudent Poster Session
https://www.stat.washington.edu/event/seminar/student-poster-session
<span>Student Poster Session</span>
<span><span lang="" about="https://www.stat.washington.edu/user/68" typeof="schema:Person" property="schema:name" datatype="" xml:lang="">trpham</span></span>
<span>Thu, 02/15/2018 - 08:58</span>
<div class="field field--name-body field--type-text-with-summary field--label-hidden field--item"><p><span><span><span>Interested in what our graduate students have been working on? Come join us for posters and presentations by the students themselves as they present their research.</span></span></span></p>
<p><span><span><span>Volunteer presenters include: </span></span></span></p>
<ul><li><span><span><span>Amrit Dhar</span></span></span></li>
<li><span><span><span>Austin Schumacher</span></span></span></li>
<li><span><span><span>Bryan Martin</span></span></span></li>
<li><span><span><span>Christopher Aicher</span></span></span></li>
<li><span><span><span>Corinne Jones</span></span></span></li>
<li><span><span><span>Hannah Director</span></span></span></li>
<li><span><span><span>Hugh Chen</span></span></span></li>
<li><span><span><span>Johnny Paige</span></span></span></li>
<li><span><span><span>Max Schneider</span></span></span></li>
<li><span><span><span>Mengjie Pan</span></span></span></li>
<li><span><span><span>Richard Guo</span></span></span></li>
<li><span><span><span>Sam Wang</span></span></span></li>
<li><span><span><span>Wesley Lee</span></span></span></li>
<li><span><span><span>Zhihang Dong</span></span></span></li>
</ul></div>
Thu, 15 Feb 2018 16:58:12 +0000trpham5007 at https://www.stat.washington.edu