Contagion effects, also known as peer effects or social influence process, have become more and more central to social science, especially with the availability of longitudinal social network data. However, contagion effects are usually difficult to identify, as they are often entangled with other factors, such as homophily in the selection process, the individual’s preference for the same social settings, etc. Methods currently available either do not solve these problems or require strong assumptions. Following Shaliziand Thomas (2011), I frame this difficulty as an omitted variable bias problem, and I propose several alternative estimation methods that have potentials to correctly identify contagion effects when there is an unobserved trait that co-determines the influence and the selection. The Monte-Carlo simulation results suggest that a latent-space adjusted estimator is especially promising. It outperforms other estimators that are traditionally used to deal with the unobserved variables, including a structural equation based estimator and an instrumental variable estimator.

We conduct a quantitative risk analysis of non-core insurance business of selling protection to financial firms against investment losses due to a shock. A static structural model is constructed, composed of a network of firms who cross-hold each other, multiple primitive assets that are vulnerable to a shock, and an insurer who resides external to the network and speculates in selling protection to the financial firms. Assume that each firm in the network is rational and able to decide how much protection to purchase to optimize its portfolio according to the meanvariance principle. As a result, the shock may impact on the insurer but indirectly through the network. More precisely, the network integration, which refers to the level of exposures of the firms to each other, aspects the way that the shock impacts on this non-core insurance business. Our study finds that the network integration and the shock play an interactive role in the insurance risk: An increase in the network integration can either reduce or amplify the impact of the shock on the insurance risk.

Bio: Dr. Qihe Tang joined the UNSW Business School as a Full Professor under the Strategic Hires and Retention Pathways (SHARP) scheme in July 2017.

After earning his Ph.D. in Statistics from the University of Science and Technology of China in 2001, he has worked at different places in the world including the University of Hong Kong (2001), the University of Amsterdam (2002-2004), the Concordia University (2004-2005), and the University of Iowa (2006-2017). At the University of Iowa, he was promoted to Full Professor in July 2012, and he was conferred the F. Wendell Miller Endowed Professorship in July 2014 in honour of his scholarly work and professional contributions.

Qihe Tang’s expertise centers on extreme value theory for insurance, finance, and quantitative risk management. Recently, he has been working on various topics newly arising from the interdisciplinary area of insurance, finance, probability, and statistics. These topics include: (1) interplay of insurance and financial risks, (2) large credit portfolio losses, and (3) modeling, measuring, and managing catastrophe risks. His research on these topics has been constantly supported by external grants.

Qihe Tang has recently been elected as an editor for Insurance: Mathematics and Economics. Currently, he is also an associate editor for the journals TEST, Applied Stochastic Models in Business and Industry, and Statistics & Probability Letters, and serves on the editorial boards of the journals Risks and Dependence Modeling. He has graduated a number of doctoral students who are now university professors all over the world.

In the weighted logrank tests such as Fleming-Harrington test and the Tarone-Ware test, certain weights are used to put more weight on early, middle or late events. The purpose is to maximize the power of the test. The optimal weight under an alternative depends on the true hazard functions of the groups being compared, and thus cannot be applied directly. We propose replacing the true hazard functions with their estimates and then using the estimated weights in a weighted logrank test. However, the resulting test does not control type I error correctly because the weights converge to 0 under the null in large samples. We then adjust the estimated optimal weights for correct type I error control while the resulting test still achieves improved power compared to existing weighted logrank tests, and it is shown to be robust in various scenarios. Extensive simulation is carried out to assess the proposed method and it is applied in several clinical studies in lung cancer.

We compute doubly truncated moments for the selection elliptical (SE) class of distributions, which includes some multivariate asymmetric versions of well-known elliptical distributions, such as, the normal, Student’s t, among others. We address the moments for doubly truncated members of this family, establishing neat formulation for high order moments as well as for its first two moments. We establish sufficient and necessary conditions for their existence. Further, we propose computational efficient methods to deal with extreme settings of the parameters, partitions with almost zero volume or no truncation. Applications and simulation studies are presented in order to illustrate the usefulness of the proposed methods.


Bio: Dr. Christian Galarza is a professor in the faculty of Mathematics and Natural Science at Escuela Superior Politécnica del Litoral (ESPOL), in Guayaquil, Ecuador. He obtained his PhD degree in Statistics in 2020, from the State University of Campinas, UNICAMP, Brazil, where he obtained his Master degree in Statistics as well. He visited the University of Connecticut from 2018 to 2019 as a research scholar. His research interests include quantile regression, linear/nonlinear mixed-effects models, EM and SAEM algorithms, zero-quantile distributions, scale mixture of skew normal distributions and censored and zero-inflated models.

Rapid developments in streaming data technologies have enabled real-time monitoring of human activity. Wearable devices, such as wrist-worn sensors that monitor gross motor activity (actigraphy), have become prevalent. An actigraph unit continually records the activity level of an individual, producing large amounts of high-resolution measurements that can be immediately downloaded and analyzed. While this type of BIG DATA includes both spatial and temporal information, we argue that the underlying process is more appropriately modeled as a stochastic evolution through time, while accounting for spatial information separately. A key challenge is the construction of valid stochastic processes over paths. We devise a spatial-temporal modeling framework for massive amounts of actigraphy data, while delivering fully model-based inference and uncertainty quantification. Building upon recent developments in scalable inference, we construct temporal processes using directed acyclic graphs (DAG) and develop optimized implementations of collapsed Markov chain Monte Carlo (MCMC) algorithms for Bayesian inference. We test and validate our methods on simulated data and subsequently apply and verify their predictive ability on an original dataset from the Physical Activity through Sustainable Transport Approaches (PASTA-LA) study conducted by UCLA’s Fielding School of Public Health.

In this article, we apply non-convex regularization methods in order to obtain stable estimation of loss development factors in insurance claims reserving. Among the non-convex regularization methods, we focus on the use of the log-adjusted absolute deviation (LAAD) penalty and provide discussion on optimization of LAAD penalized regression model, which we prove to converge with a coordinate descent algorithm under mild conditions. This has the advantage of obtaining a consistent estimator for the regression coefficients while allowing for the variable selection, which is linked to the stable estimation of loss development factors. We calibrate our proposed model using a multi-line insurance dataset from a property and casualty insurer where we observed reported aggregate loss along accident years and development periods. When compared to other regression models, our LAAD penalized regression model provides very promising results.

Bio: Dr. Himchan Jeong is an Assistant Professor in the Department of Statistics and Actuarial Science at Simon Fraser University, Canada. He is a Fellow of the Society of Actuaries (SOA) and holds a Ph.D. from the University of Connecticut. He has been actively involved in teaching and conducting research in actuarial science for several years. In recognition for his academic achievements and excellence, he has been awarded the James C. Hickman Scholarship from SOA recently in 2018-2020.
His current research interest is predictive modeling for ratemaking and reserving of property and casualty insurance.

Consider the subgraph sampling model, where we observe a random subgraph of a given (possibly non random) large graph $G_n$, by choosing vertices of $G_n$ independently at random with probability $p_n$. In this setting, we study the question of estimating the number of copies $N(H,G_n)$ of a fixed motif/small graph (think of $H$ as edges, two stars, triangles) in the big graph $G_n$. We derive necessary and sufficient conditions for the consistency and the asymptotic normality of a natural Horvitz-Thompson (HT) type estimator. 

As it turns out, the asymptotic normality of the HT estimator exhibits an interesting fourth-moment phenomenon, which asserts that the HT estimator (appropriately centered and rescaled) converges in distribution to the standard normal whenever its fourth-moment converges to 3. We apply our results to several natural graph ensembles, such as sparse graphs with bounded degree, Erdős-Renyi random graphs, random regular graphs, and dense graphons.

When studying real-world data, such as the electronic health records (EHR), data collected from multiple sites enable analyses to be conducted with larger sample sizes and increased generalizability. Many distributed algorithms have been proposed and applied to data integration across multiple sites, by requiring only aggregated data (AD) rather than individual patient data (IPD) for the privacy-preserving reason. However, it is often been ignored that data from different sites are heterogeneous. In this talk, two novel privacy-preserving distributed algorithms (PDA) are introduced for heterogeneous data integration, namely the distributed proportional likelihood ratio model (DPLR) and distributed linear mixed model (DLMM). Both algorithms require each site to communicate some AD once, but obtain estimates that are close or identical to that of pooled analyses. The two algorithms have been applied to study the avoidable hospitalization of pediatric patients across multiple care sites within the Children’s Hospital of Philadelphia (CHOP), and the LOS of COVID-19 hospitalization across multiple databases. 

Bio: Dr. Chongliang Luo is currently a postdoctoral fellow at the Department of Biostatistics, Epidemiology and Informatics, University of Pennsylvania. He graduated with a Ph.D. in Statistics at the University of Connecticut. His research interests focus on statistical methodology on data integration, including multi-view data integration, statistical methods on meta-analysis and privacy-preserving distributed statistical learning.

We investigate non-uniform negative sampling and estimation for imbalanced data. Imbalanced data is ubiquitous in binary response problems, where the number of events (positive class) is significantly smaller than the number of nonevents (negative class). For this scenario, the available information is at the scale of the number of positive instances instead of the full data sample size, and subsampling the negative class is a common practice to reduce the computation and/or data collection costs. We first derive the asymptotic distribution of a general inverse probability weighted (IPW) estimator and minimize its variance to obtain optimal sampling probabilities. To further improve the estimation efficiency, we propose a likelihood-based estimator by correcting log odds for the sampled data. The improved estimator is applicable to all binary response models with any forms of sampling probabilities, and its asymptotic variance is smaller than that of the IPW estimator. We validate our approach on simulated data as well as a real click-through rate dataset with more than 0.3 trillion instances, collected over a period of a month. Both theoretical and empirical results demonstrate the effectiveness of our method. 

Bio: HaiYing Wang is an Assistant Professor in the Department of Statistics at the University of Connecticut. He was an Assistant Professor in the Department of Mathematics and Statistics at the University of New Hampshire from 2013 to 2017. He obtained his Ph.D. from the Department of Statistics at the University of Missouri in 2013, and his M.S. from the Academy of Mathematics and Systems Science, Chinese Academy of Sciences in 2006. His research interests include informative subdata selection for big data, model selection, model averaging, measurement error models, and semi-parametric regression.

http://haiying-wang.uconn.edu

We study Granger causality testing for high-dimensional time series using regularized regressions. To perform proper inference, we rely on heteroskedasticity and autocorrelation consistent (HAC) estimation of the asymptotic variance and develop the inferential theory in the high-dimensional setting. To recognize the time series data structures we focus on the sparse-group LASSO estimator, which includes the LASSO and the group LASSO as special cases. We establish the debiased central limit theorem for low dimensional groups of regression coefficients and study the HAC estimator of the long-run variance based on the sparse-group LASSO residuals. This leads to valid time series inference for individual regression coefficients as well as groups, including Granger causality tests. The treatment relies on a new Fuk-Nagaev inequality for a class of $\tau$-mixing processes with heavier than Gaussian tails, which is of independent interest. In an empirical application, we study the Granger causal relationship between the VIX and financial news.

We develop the scale transformed power prior for settings where historical and current data involve different data types, such as binary and continuous data, respectively. This situation arises often in clinical trials, for example, when historical data involve binary responses and the current data involve time-to-event or some other type of continuous or discrete outcome. The power prior proposed by Ibrahim and Chen (2000) does not address the issue of different data types. Herein, we develop a current type of power prior, which we call the scale transformed power prior (straPP). The straPP is constructed by transforming the power prior for the historical data by rescaling the parameter using a function of the Fisher information matrices for the historical and current data models, thereby shifting the scale of the parameter vector from that of the historical to that of the current data. Examples are presented to motivate the need for a scale transformation and simulation studies are presented to illustrate the performance advantages of the straPP over the power prior and other informative and non-informative priors. A real dataset from a clinical trial undertaken to study a novel transitional care model for stroke survivors is used to illustrate the methodology.