Shiny Essentials (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): Mine Cetinkaya-Rundel, Duke University
Shiny is an R package that makes it easy to build interactive web apps straight from R. You can host stand-alone apps on a webpage or embed them in R Markdown documents or build dashboards. This short course will introduce you to building web applications and dashboards with Shiny, reactive programming, and customizing and deploying your apps for others to use. Please bring a laptop with you.
Master the Tidyverse: An Introduction to R for Data Science (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): Garrett Grolemund, RStudio Inc.
This two-day workshop covers the new book "R for Data Science" from Hadley Wickham and Garrett Grolemund. The workshop provides a comprehensive overview of what is now called the Tidyverse, a core set of R packages that are essential to Data Science. We will visualize, transform, and model data in R and work with date-times, character strings, and untidy data formats. Along the way, you will learn to use the brightest stars in the tidyverse: the ggplot2, dplyr, tidyr, readr, purrr and tibble packages along with stringr, lubridate, hms, and forcats. Our objective will be to learn to use R to do the main tasks of data analysis efficiently. A basic knowledge of R syntax is assumed. This course won the "Excellence-in-CE Award" at JSM 2017 and was invited to return for JSM 2018.
Joint Modeling of Longitudinal and Time-to-Event Data (ADDED FEE) — Professional Development Continuing Education Course
ASA, Biometrics Section
Instructor(s): Robert Elashoff, UCLA; Gang Li, UCLA; Ning Li, UCLA
Longitudinal analysis and time-to-event data analysis are among the fastest expanding areas of statistics and biostatistics in the past three decades. In recent years, these two seemingly different areas of statistics have crossed with the rapidly growing interest in development of joint models for longitudinal and time-to-event data to address challenging issues that cannot be properly handled using standard methods within each area. This course aims to a give a systematic introduction and review of state-of-the-art statistical methodology developed in recent years for joint models. We will provide motivating examples and an overview of statistical modeling and concepts that are fundamental to understand joint models, discuss several main areas in which joint models have been developed to address important scientific questions and issues, including non-ignorable missing data in longitudinal analysis, event time models with intermittently measured time-dependent covariates, longitudinal studies with informative observation times, joint models for competing risks event times, and some further topics. The last section will give the audience hands-on experience of analyzing data using joint models. Examples will be illustrated by computer programs in R. The course concludes with a self-practice session.
Bayesian Thinking: Fundamentals, Computation, and Multilevel Modeling (ADDED FEE) — Professional Development Continuing Education Course
ASA, Section on Bayesian Statistical Science
Instructor(s): Jim Albert, Bowling Green State University
The basic tenets of Bayesian thinking are introduced, including construction of priors, summarization of the posterior to perform inferences, and the use of prediction distributions for prediction and model checking. There will be a focus on Bayesian regression for continuous and categorical response data. Bayesian multilevel models are introduced as a flexible way of modeling regressions over groups. The use of R in Bayesian computation is described, including the programming of the posterior distribution and the use of different R tools to summarize the posterior. Special focus will be on the application of Markov chain Monte Carlo algorithms and diagnostic methods to assess convergence of the algorithms. The LearnBayes and rethinking R packages are used to illustrate MCMC fitting by the use of gibbs sampling and Metropolis algorithms. Larger Bayesian models will be fit using JAGS and Stan and the accompanying rjags and rstan packages. Some familiarity of the participant with the R statistical language would be helpful.
Propensity Score Methods: Concepts and Applications (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): Haiyan Bai, University of Central Florida; Wei Pan, Duke University
Observational studies are often conducted to discover real world evidence of comparative effectiveness using real world data, but selection bias in real world data poses threats to the validity of real world evidence. Propensity score methods (PSMs) have been increasingly used in observational studies as a means of reducing selection bias. This course will introduce concepts, applications, and issues of PSMs, including matching, stratification, and weighting. We will also discuss when and how to apply PSMs in observational studies using real world data. Through lectures on the concepts of PSMs and hands-on activities for the use of statistical programs in R and SAS, this course will benefit faculty members, graduate students, and applied researchers improving the quality of observational studies. Instructions for downloading and installing related statistical programs and examples of real world data will be provided to participants in advance through a course website, and additional handouts will be also made available in class. No prior knowledge of PSMs is required. However, an understanding of basic research design and statistics is preferred. Participants are encouraged to bring their own laptops for hands-on activities during which participants are also welcomed to work on their own real world data.
Advanced R (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): Charlotte Wickham, Oregon State University
Move from using other peoples' functions to writing your own! This course focuses on some of the key programming techniques in R that will move you to the next level. You'll get the most from this course if you have some R programming experience already: you've written a few R functions and are comfortable with R's basic data structures (vectors, lists and data frames). We'll start with a review of some R fundamentals before jumping into writing functions in R. You'll learn some strategies for getting started, and making your functions easy for others to use. Once you've mastered writing functions, you learn about the ways functions can be used in R, like functions that write other functions and functions as arguments to functions - key elements of functional programming. You'll also learn about `purrr`, a package that enhances R's functional programming toolkit. Finally, you'll learn about tidy evaluation: a framework for creating domain specific languages. Tidy evaluation makes it easy for you to program with functions that use it (e.g. functions in the tidyverse like `dplyr::filter()` and `tidyr::spread()`). This course is taught by Charlotte Wickham, Assistant Professor at Oregon State University, independent trainer and DataCamp instructor.
Introduction to the Design of Cluster Randomized Trials (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): Lawrence Moulton, Dept. International Health
This course will cover the main concepts and methods in the design of trials in which the unit of randomization is a group of participants. There has been a steep increase in employing this kind of study to answer research questions in situations where a treatment/intervention must be applied at a group level to: 1) avoid contamination; 2: estimate parameters that are functions of within-group dynamics; or 3: meet logistical exigencies. The course will consist of an overview of key design features of cluster randomized trials, including parameters of interest, cluster formation, highly restricted randomization, and sample size requirements. The use of matching, stratification, and phased implementation (including stepped wedge) techniques will be presented. Many of the examples will involve infectious disease interventions in geographically-randomized settings, but most of the principles are relevant to any group-randomized scenario. The primary source material may be found in: Hayes, R.J. and Moulton, L.H. (2017) Cluster Randomised Trials, 2nd Edition, Chapman & Hall.
Technical aspects will be minimal (bachelors/masters level), although familiarity with the basics of clinical trial design is strongly recommended. The use of software routines (in Stata, SAS, or Excel) will be touched on, but no prior experience with these is assumed.
Preparing Statisticians for Leadership: How to See the Big Picture and Have More Influence (ADDED FEE) — Professional Development Personal Skills Development
ASA
Instructor(s): Gary Sullivan; Bonnie LaFleur, HTG Molecular Diagnostics, Inc.
What is leadership? Much has been written and discussed within the statistics profession in the last few years on the topic and its importance in advancing our profession. This course provides an understanding of leadership and how statisticians can improve and demonstrate leadership to affect their organizations. It features leaders from all sectors of statistics speaking about their personal journeys and provides guidance on personal leadership development with a focus on the larger organizational/business view and influence. Course participants work with their colleagues to discuss and resolve leadership situations that statisticians face. Participants will come away with a plan for developing their own leadership and connect with a network of statisticians who can help them move forward on their leadership journey.
Statistical and Computational Methods and Software for Microbiome and Metagenomics and Applications (ADDED FEE) — Professional Development Continuing Education Course
ASA, Section on Statistics in Genomics and Genetics
Instructor(s): Curtis Huttenhower, Harvard University; Hongzhe Li, University of Pennsylvania
The microbial organisms in and on human body constitute the human microbiome, of which the largest collection of microbes resides in the gut. The gut microbiome plays an important role in human health and disease. Research in the last decades have clearly shown that perturbation of the microbial ecosystem could be responsible for many complex . High-throughput sequencing technologies make it possible to interrogate the metagenomics of all microbes in human gut, which allows us to characterize composition of both microbe genes and species, to quantify the dynamics of a microbial ecosystem and to understand their role in human diseases. Analysis of such microbiome and metagenomics data is however challenging due to complexity of the communities and high dimensionality of the data. This course will introduce the statistical and computational tools for functional metagenomics, including methods for accurately estimating the microbial composition, genes and functional group composition of a microbiome, methods for identifying disease-associated bacterial taxa based on high dimensional compositional data and their mediating. Software demos will be also provided so that students will leave the course with the tools necessary to perform common analyses of microbiome and metagenomics data. Detailed analysis of the HMP data will be presented.
Preparing Statisticians for Leadership: How to See the Big Picture and Have More Influence (ADDED FEE) — Professional Development Personal Skills Development
ASA
Instructor(s): Gary Sullivan; Bonnie LaFleur, HTG Molecular Diagnostics, Inc.
What is leadership? Much has been written and discussed within the statistics profession in the last few years on the topic and its importance in advancing our profession. This course provides an understanding of leadership and how statisticians can improve and demonstrate leadership to affect their organizations. It features leaders from all sectors of statistics speaking about their personal journeys and provides guidance on personal leadership development with a focus on the larger organizational/business view and influence. Course participants work with their colleagues to discuss and resolve leadership situations that statisticians face. Participants will come away with a plan for developing their own leadership and connect with a network of statisticians who can help them move forward on their leadership journey.
Master the Tidyverse: An Introduction to R for Data Science (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): Garrett Grolemund, RStudio Inc.
This two-day workshop covers the new book "R for Data Science" from Hadley Wickham and Garrett Grolemund. The workshop provides a comprehensive overview of what is now called the Tidyverse, a core set of R packages that are essential to Data Science. We will visualize, transform, and model data in R and work with date-times, character strings, and untidy data formats. Along the way, you will learn to use the brightest stars in the tidyverse: the ggplot2, dplyr, tidyr, readr, purrr and tibble packages along with stringr, lubridate, hms, and forcats. Our objective will be to learn to use R to do the main tasks of data analysis efficiently. A basic knowledge of R syntax is assumed.
Regression Modeling Strategies (ADDED FEE) — Professional Development Continuing Education Course
ASA, Biometrics Section
Instructor(s): Frank Harrell, Vanderbilt University, Dept of Biostatistics
All standard regression models have assumptions that must be verified for the model to have power to test hypotheses and for it to be able to predict accurately. Of the principal assumptions (linearity, additivity, distributional), this course will emphasize methods for assessing and satisfying the first two. Practical but powerful tools are presented for validating model assumptions and presenting model results. This course provides methods for estimating the shape of the relationship between
predictors and response using the widely applicable method of augmenting the design matrix using restricted cubic splines. Even when assumptions are satisfied, over-fitting can ruin a model's predictive ability for future observations. Methods for data reduction will be introduced to deal with the common case where the number of potential predictors is large in comparison with the number of observations. Methods of model validation (bootstrap and cross-validation) will be covered, as will auxiliary topics such as modeling interaction surfaces, variable selection, overly influential observations, collinearity, and shrinkage, and a brief introduction to the R rms package for handling these problems. The methods covered will apply to almost any regression model, including ordinary least squares, logistic regression models, ordinal regression, quantile regression, longitudinal data analysis, and survival models.
Art and Practice of Regression Trees and Forests (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): Wei-Yin Loh, University of Wisconsin
Regression tree and forest methods have greatly improved in the last decade. Their ease of use, prediction accuracy, execution speed, and interpretability make them essential tools for machine learning and data analysis. The course teaches how to use the tools effectively and efficiently in practice. It follows an example-focused style, with each example chosen to illustrate particular weaknesses of traditional solutions and to show how tree methods overcome them and yield new insights. Examples include a large consumer survey with hundreds of variables and substantial amounts of missing values; cancer and diabetes randomized trials with censored and longitudinal responses for precision medicine; and observational studies of high-school dropouts and Alzheimer's patients. Learning highlights are (1) how trees deal with missing values without requiring imputation, (2) how importance scores help with variable selection, and (3) how to perform post-selection inference with the bootstrap. To encourage hands-on training, the presentation is interwoven with live demos of free software. No commercial software is required. Specific algorithmic techniques are discussed where appropriate but no systematic presentation of entire algorithms is given. Attendees should have experience with linear and logistic regression. Instructions for software and dataset downloads will be given in advance.
Practical Considerations for Bayesian and Frequentist Adaptive Clinical Trials (ADDED FEE) — Professional Development Continuing Education Course
ASA, Section on Bayesian Statistical Science
Instructor(s): Frank Bretz, Novartis Pharma AG; Byron Jones, Novartis Pharma AG; Peter Müller, University of Texas Austin
Clinical trials play a critical role in pharmaceutical drug development. New trial designs often depend on historical data, which, however, may not be accurate for the current study due to changes in study populations, patient heterogeneity, or different medical facilities. As a result, the original plan and study design may need to be adjusted or even altered to accommodate new findings and unexpected interim results. Through carefully thought-out and planned adaptation, the right dose can be identified faster, patients can be treated more effectively, and treatment effects evaluated more efficiently.
This one-day short course will introduce various adaptive methods for Phase I to Phase III clinical trials using both, frequentist and Bayesian methods. Accordingly, different types of adaptive designs will be introduced and practical
considerations will be illustrated with case studies. Types of adaptive clinical trial designs covered in this course include dose escalation/de-escalation and dose insertion based on observed data, adaptive dose finding studies using optimal designs to allocate new cohorts of patients based on the accumulated evidence, blinded and unblinded sample size reestimation as well as adaptive designs for confirmatory trials with treatment or population selection at interim.
Data Science Workflows Using R and Spark (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): E. James Harner, West Virginia University
This short course covers the data science process using R as a programming language and Spark as a big-data platform. Powerful workflows are developed for data extraction, data transformation and tidying, data modeling, and data visualization. The course is taught using a Dockerized virtual cluster with containers for R and RStudio, PostgreSQL, Hadoop, Spark and various NoSQL databases. The interface to the computational environment is a modern web browser, whether the Docker deployment is local or remote. During the course R-based and bash illustrations show how data is transported using REST APIs, sockets, etc. into persistent data stores such as the Hadoop Distributed File System (HDFS), NoSQL databases, relational databases and in some cases sent directly to Spark's real-time compute engine. Workflows using dplyr verbs are used for data manipulation within R, within relational databases (PostgreSQL), and then within Spark using sparklyr. These data-based workflows extend into machine learning algorithms, model evaluation, and data visualization using sparklyr and ggplot2.
The machine learning algorithms are taught using Spark's distributed computational engine on data stored in HDFS or distributed in-memory. Concepts of data locality and methods of avoiding data shuffling are discussed. The supervised techniques include linear regression, logistic regression, generalized linear regression, decision trees, gradient-boosted trees, and random forests. Feature selection is done primarily by regularization and models are evaluated using various metrics. Unsupervised techniques include k-means clustering and dimension reduction. TensorFlow for deep learning will be introduced. Big-data architectures are discussed including the Docker containers used for building the course infrastructure called rspark. See: https://github.com/jharner/rspark
The Docker containers can be run on the desktop, run using vagrant, or deployed to Amazon Web Services (AWS). As a result, students will have access to a full big-data computing platform and extensive course content.
JSM Presentation Skills Workshop (FREE - NO REGISTRATION REQUIRED) — Professional Development Personal Skills Development
ASA
Organizer(s): Brian Wiens, Allergan
Instructor(s): Scott Berry, Berry Consultants; Richard De Veaux, Williams College; William Li, Shanghai Advanced Institute of Finance
A panel of experienced and award-winning presenters will share advice about speaking at JSM. Topics will include the following: Engaging the audience, Effective practice techniques, organizing your talk, visual aids, answering audience questions, and speaking in a language other than your native tongue. All presenters are welcome, though first-time speakers at JSM are especially encouraged to attend. Also, anyone who is considering a future JSM presentation, or just wants to hear more about the art of scientific speaking, is welcome.
Career Development Panel (FREE - NO REGISTRATION REQUIRED) — Professional Development Personal Skills Development
ASA
Equality in Career Development: Advocating and Embodying Equality in Career Development and Leadership: Statisticians may become leaders within their organizations as they progress through their careers. As such, it can be important to think about developing good leadership qualities early on rather than when promoted into a leadership role. Promoting equality in your organization is one leadership area that can be developed throughout one's career. Many factors affect a statistician's ability to lead effectively in the workplace, but as a statistical leader, confidence, overcoming feelings of dissimilarity, and clear communication are paramount, especially since statisticians are usually working with professionals in other fields who are already professionally dissimilar to them. The panel will be composed of a diverse group of statistical leaders. Panelists will share examples and advice for advocating and embodying equality in leadership as statisticians and as leaders in their organizations. The panelists will also discuss their personal stories of successful leadership tools in environments where they felt dissimilar, how they identified and fostered these skills throughout their career, advocacy of equality and/or overcoming adversity. They will also comment on how the organizations they've worked for do handle or should handle equality in leadership, for example, through zero-tolerance policies, diversity summits, rules for conduct in the workplace.
Adaptive Survey Design (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): James Wagner, University of Michigan
Many statistical agencies and survey organizations are looking for design options that control costs and errors. This situation has led to a growing interest in adaptive survey designs. Adaptive survey designs are based on the rationale that any population is both heterogeneous in its response and answering behavior to surveys and in its costs to be recruited and interviewed. Different survey design features may be effective for different members of the population. Adaptive survey designs acknowledge these differences by allowing differentiation of survey design features for different population subgroups based on auxiliary data about the sample; the auxiliary data is linked from frame data, registry data or paradata. The strata receive different treatments. This course will focus on practical guidance for building adaptive survey designs, including identification of strata, choice of strategies, and optimization of design features across strata.
Meta-Analysis for Biopharmaceutical and Public Health Research Using SAS (ADDED FEE) — Professional Development Continuing Education Course
ASA, Biopharmaceutical Section, Section on Teaching of Statistics in the Health Sciences
Instructor(s): Michael LaValley, Boston University School of Public Health; Ludovic Trinquart, Boston University School of Public Health
Meta-analysis is the gold standard statistical approach to combine the results of multiple studies and to examine sources of heterogeneity and potential biases. This course will review fixed-effect and random-effects models, including the newly-recommended approaches, that underlie the combination of study results in meta-analysis; the use of study-level predictors in meta-regression; novel limit meta-analysis models to adjust for small-study effects and related reporting biases; and the synthesis of individual participant data. Throughout the course, participants will apply each model by using the SAS/STAT software and produce high-quality graphs, with original macros developed by the instructors. The target audience includes biostatisticians, data analysts, and quantitative researchers from academia, the pharmaceutical industry, the FDA and other government institutions with a basic knowledge of study design and regression modeling.
Effective Presentations for Statisticians: Success = (PD)2 (ADDED FEE) — Professional Development Personal Skills Development
ASA
Instructor(s): Jennifer Van Mullekom, Virginia Tech
Public speaking is the number-one fear in America, and yet being able to do so is absolutely critical for success in business settings. Statisticians must be able to effectively convey their ideas to clients, collaborators, and decision-makers.
Presenting in the modern world is even more daunting when speakers have the opportunity to employ slideware, videos, and live demonstrations. Unfortunately, university coursework and professional development programs are often not targeted toward sharpening these skills. This short course, developed and taught by statisticians, will provide an opportunity to learn how to employ different methods and tools in the phases of the Success = (PD)2 framework.
The material covered is geared toward scientific presentations and based on the works of Garr Reynolds and Michael Alley, among others. The course will emphasize the importance of stepping away from the computer to prepare an effective message aimed at your core point guided with a series of questions and tips.
The design phase emphasizes the importance of structure, streamlining, and good graphic design accompanied by a series of checklists. Of course, practice makes perfect, so we cannot skip this step.
Finally, engaging the audience and effectively using the room and equipment is covered in the deliver phase and is complemented with a handy list of dos and don'ts. No matter where you are in your journey for presentation success, improvement is always possible.
We look forward to seeing you in this valuable class, where you can hone your skills! Be prepared for an active class full of discussion and group exercises.
Instructor(s): Michael Daniels, University of Florida; Jason Roy, University of Pennsylvania
Bayesian nonparametric (BNP) methods can be used to flexibly model joint or conditional distributions, as well as functional relationships. These methods, along with causal assumptions, can be used with the g-formula for inference about causal effects. This general approach to causal inference has several possible advantages over popular semiparametric methods, including efficiency gains, the ease of causal inference on any functionals of the distribution of potential outcomes, the use of prior information, and capturing uncertainty about causal assumption via informative prior distributions. In this short course we review BNP methods and illustrate their use for causal inference in the setting of point treatments, dynamic (longitudinal) treatments, and mediation. We present several data examples and discuss software implementation using R. The R code and/or packages used to run the data examples will be provided to the attendees at a specific github site.
Nonparametric Regression and Classification for Modern Data Scientists (ADDED FEE) — Professional Development Continuing Education Course
ASA, Section on Statistical Learning and Data Science
Instructor(s): David Banks, Duke University; Margaret Johnson, SAMSI
This course surveys fundamental concepts in modern data science. It emphasizes nonparametric regression and classification,
with sparsity, regularization, and the Curse of Dimensionality being recurring themes. Specific topics include:
(1) nonparametric regression, including the backfitting algorithm, with the bootstrap and cross-validation as associated tools, (2) the Lasso, elastic net, and LARS, with the Hoff algorithm for solutions when the penalty function is Lq for 0 < q < 1, (3) the p >> n problem, with a survey of key results from Donoho and Tanner, Candes and Tao, and Wainwright, (4) the median model of Berger and Barbieri, (5) comparison of geometric, algorithmic and probabilistic classification methodology, including nearest-neighbor,support vector machine, and Random Forests techniques, (6) improvement of classification techniques through ensembles and forward stagewise learning, such such as bagging, stacking and boosting, (7) topic modeling, using Latent Dirichlet Allocation. Most ideas will be illustrated through an application to a data set.
Julia Meets Mendel: Algorithms and Software for Modern Genomic Data Analysis (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): Janet Sinsheimer, UCLA; Eric Sobel, UCLA; Hua Zhou, UCLA
Challenges in statistical genomics and precision medicine are enormous. Datasets are becoming bigger and more varied, demanding complex data structures and integration across multiple biological scales. Analysis pipelines juggle many programs, implemented in different languages, running on different platforms, and requiring different I/O formats. This heterogeneity erects barriers to communication, data exchange, data visualization, biological insight and replication of results. Statisticians spend inordinate time coding/debugging low-level languages instead of creating better methods and interpreting results. The benefits of parallel and distributed computing are largely ignored.
The time is ripe for better statistical genomic computing approaches. This short course reviews current statistical genomics problems and introduces efficient computational methods to (1) enable interactive and reproducible analyses with visualization of results, (2) allow integration of varied genetic data, (3) embrace parallel, distributed and cloud computing, (4) scale to big data, and (5) facilitate communication between statisticians and their biomedical collaborators. We present statistical genomic examples and offer participants hands on coding exercises in Julia as part of the OpenMendel project (https://openmendel.github.io). Julia is a new open source programming language with a more flexible design and superior speed over R and Python. R and Matlab users quickly adapt to Julia.
Analysis of Continuous and Categorical Method Comparison Data (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): Pankaj Choudhary, University of Texas at Dallas; Haikady Nagaraja, Ohio State University
Hundreds of method comparison studies for evaluating agreement between two or more methods of measuring a continuous or categorical response are published each year in biomedical disciplines. This course will discuss statistical methodologies for analysis of such studies. For continuous response, we will focus on a new approach that involves modeling of data by a mixed-effects model and performing inference on measures of agreement, such as limits of agreement, concordance correlation, and total deviation index. Besides providing a more informative analysis than the popular approach of Bland and Altman (1986, Lancet) that has over 25000 citations, this approach offers a unified framework for analyzing a variety of data, including paired measurements, repeated measurements, and longitudinal data and data from multiple methods, which will be specifically considered in the course. For categorical response, we will focus on kappa coefficients and related measures. A detailed case-study will be presented for illustrating the analysis of each data type. Relevant R software code will be provided. The course is based on a new Wiley monograph, Measuring Agreement, by the instructors, and presumes knowledge of the basics of statistical inference, regression modeling, and R. Some familiarity with mixed-effects models is beneficial but not necessary.
Prediction in Event-Based Clinical Trials (ADDED FEE) — Professional Development Continuing Education Course
ASA, Biometrics Section
Instructor(s): Daniel Heitjan, Southern Methodist University; Gui-shuang Ying, University of Pennsylvania
Randomized clinical trials often include planned interim analyses, at which external reviewers assess the accumulated data to determine whether the study should continue. With time-to-event endpoints, it is often desirable to schedule the interim analyses at the times of occurrence of specified landmark events, such as the 100th event, the 200th event, and so on. It can be worthwhile to predict the times of such events, together with other trial outcomes, as an aid to real-time logistical planning.
Traditional prediction methods use data only from previous trials and give inaccurate projections if, as often happens, historical enrollment or event rates differ from those in the current trial. With modern data management systems we can create accurate and complete study databases in real time, making it possible to use the accumulating data from the trial itself to make predictions about its future. Over the last several years the presenters have developed a suite of statistical methods for real-time prediction of the future course of a clinical trial. In this short course we will describe these methods and train potential users in their application.
Statistical Methods for Single-Cell RNA-Seq Analysis (ADDED FEE) — Professional Development Continuing Education Course
ASA, Section on Statistics in Genomics and Genetics
Instructor(s): Rhonda Bacher, University of Wisconsin - Madison; Christina Kendziorski, University of Wisconsin - Madison; Mingyao Li, University of Pennsylvania; Nancy Zhang, University of Pennsylvania
Single-cell RNA-sequencing (scRNA-seq) has emerged as a revolutionary tool that allows us to address scientific questions that were elusive just a few years ago. With the advantages of scRNA-seq come statistical challenges that are just beginning to be addressed. In this course, we will review the computational and statistical methods available for the design and analysis of scRNA-seq experiments including methods for quality control, normalization, accounting for technical noise, gene expression estimation and recovery, allele specific expression estimation, sub-population identification, pseudotemporal ordering and inference, and identification of differential distributions. Advantages and disadvantages of approaches in various settings will be discussed. Software demos will be also provided so that students will leave the course with the tools necessary to perform common analyses of single-cell RNA-seq data.
Public Lecture: Born on Friday the Thirteenth: The Curious World of Probabilities — Invited Special Presentation
ASA
Free and Open to all:
This talk will use randomness and probability to answer such questions as: Just how unlikely is it to win a lottery jackpot? If you flip 100 coins, how close will the number of heads be to 50? How many dying patients must be saved to demonstrate the effectiveness of a new medical drug? Why do strange coincidences occur so often? How accurate are opinion polls? How did statistics help to expose the Ontario Lottery Retailer Scandal? Should parents be convicted of murder if two of their babies die without apparent cause? Can statistics explain luck and superstition? Why do casinos always make money, even though gamblers sometimes win and sometimes lose? And how is all of this related to Monte Carlo algorithms, an extremely popular and effective method for scientific computing? No mathematical background is required to attend.
More information
on this session.
Practical Hierarchical Bayesian Modeling (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): Fang Chen, SAS Institute Inc.
This half-day course reviews the basic concepts of Bayesian hierarchical models and focuses on using software to fit multilevel models (including random-effects models). The objectives are to familiarize statistical programmers and practitioners with the essentials of the Bayesian paradigm in the area of conducting hierarchical modeling. The course places significant emphasis on equipping attendees with computational tools through a series of worked-out examples that demonstrate sound practices for fitting Bayesian hierarchical models and conducting inferences.
The first part of the course reviews Bayesian hierarchical modeling, including concepts such as components of a multilevel model, exchangeability, group-specific inferences, choice and impact of prior distributions, and concepts in estimation and prediction.
The second part of the course takes an applied approach, illustrating the Bayesian treatment of a wide range of hierarchical models by using software, with code explained in detail. Topics include linear, generalized linear, and nonlinear random-effects models; nested and non-nested models; latent variable models; models that involve temporal and spatial structures; and meta-analysis applications. Statistical topics discussed include the selection of prior distributions and their implications, sensitivity analysis, prediction, model comparison, and general inferences. The examples are done using SAS (PROC MCMC), with a strong focus on technical details.
Attendees should have a background equivalent to an MS in applied statistics. Previous exposure to Bayesian methods is useful but not required. Familiarity with material at the level of the textbook Probability and Statistics, by DeGroot and Schervish (Addison Wesley), is appropriate.
CANCELED: Health Care Analytics in the Presence of 'Big Data' (ADDED FEE) — Professional Development Continuing Education Course
ASA, Biometrics Section
Instructor(s): Evan Carey, University of Colorado School of Public Health
The phrase "big data" has become widespread, but what does it mean for the practicing healthcare analyst? How does the presence of large dimensional data impact the actual workflow of conducting analytics in health care/health policy? In this course, participants will gain experience using cutting edge software tools for big data analysis, with a focus on Python and Apache Spark. We will begin with an overview of the challenges to making inference in the presence of high dimensional data. This will lead to a discussion of recent software solutions in this space. Through instructor led examples, we will next discuss and demonstrate the efficiency of various analytic frameworks. We will differentiate online learning approaches from distributed optimization approaches. Various examples of dimensionality reduction for data in the pre-modeling environment will be covered. We will contrast traditional serial optimization approaches (such as Newton Raphson) with parallel optimization approaches (such as stochastic gradient descent). Students will be provided with code to run all models presented at the workshop, thus no experience in these languages is required. All software used will be open source; students will be able to set up their computing environment prior to the workshop.
Scientific Communication: How to Write an Op-Ed (ADDED FEE) — Professional Development Personal Skills Development
ASA
Instructor(s): Trevor Butterworth, Sense About Science USA; Elisabeth Eaves
An opinion column can be a powerful way to address the public on an important issue of the day-and influence policy. But we all have opinions, and competition to get published in the media is intense. So where do you start?
In this course, we will look at the practical realities of writing opinion columns on statistics (and science in general). How do you break down a complex issue to its core elements and build that into an argument for a general reader? How do you structure a column? What should you leave out? What voice or style should you write in? How do you pitch to an editor?
Elisabeth Eaves is a contributing editor to the Bulletin of the Atomic Scientists and former opinion editor at The Daily, Forbes, and Wall Street Journal. Trevor Butterworth is the executive director of Sense About Science USA.
Statistical Machine Learning for Biomedical Data (ADDED FEE) — Professional Development Continuing Education Course
ASA, Biometrics Section
Instructor(s): Noah Simon, University of Washington
We will present a number of supervised learning methods that can be applied to Biomedical Big Data: In particular we will cover penalized approaches to regression and classification; as well as support vector machines, and tree-based methods. We will consider the analysis of "high-dimensional Omics" data sets. These data are typically characterized by a huge number of molecular measurements (such as genes) and a relatively small number of samples (such as patients). In addition, we will discuss the use of these tools in the development of prognostic and predictive biomarkers. Each topic will be illustrated with examples, both of well-done and poorly-done analyses. The example analyses will be conducted using state-of-the-art packages in R (including "e1071", "rpart", "gbm" and "glmnet"). Throughout the course, we will focus on common pitfalls in the supervised analysis of Biomedical Big Data and how to avoid them. The course will include interactive discussions/"challenge questions", to help participants actively engage with applying these tools in biomedical scenarios. This course assumes some previous exposure to linear regression, statistical hypothesis testing and R.
Analysis of Clinical Trials: Theory and Applications (ADDED FEE) — Professional Development Continuing Education Course
ASA, Biopharmaceutical Section
Instructor(s): Alex Dmitrienko, Mediana Inc; Jeff Maca, Quintiles IMS; Devan V Mehrotra, Merck & Co., Inc.
The course covers seven important topics that commonly face statisticians and research scientists conducting clinical research: stratified trials, longitudinal trials with dropouts, time-to-event trials with small sample sizes, crossover trials, pharmacogenomics studies for personalized medicine, multiple comparisons, and interim decision making with adaptive designs.
The course offers a well-balanced mix of theory and applications. It presents practical advice from experts and discusses regulatory considerations. The discussed statistical methods will be implemented using SAS and R software. Clinical trial examples will be used to illustrate the statistical methods.
The course is designed for statisticians working in the pharmaceutical or biotechnology industries as well as contract research organizations. It is equally beneficial to statisticians working in institutions that deliver health care, government branches that conduct health-care related research and academics interested in learning about contemporary statistical topics in clinical drug development. The attendees are required to have a basic knowledge of clinical trials. Familiarity with drug development is highly desirable, but not necessary.
This course was taught at JSM 2005-2017 and received the Excellence in Continuing Education Award in 2005.
Deep Learning, Prediction, and Validation: Innovations in Statistical Modeling and Applications to Medical/Health Big Data (ADDED FEE) — Professional Development Continuing Education Course
We begin with a review of (a) recent advances in computer vision and deep learning and how it links with statistical/machine learning, (b) the underlying statistical theories of convolutional neural networks, gradient descent, graphical models, and hidden Markov random fields, and (c) AI applications to medical imaging and automated analysis of electronic medical and health data. Whereas high-performance computing and advanced programming have overcome the computational hurdles in the analysis of "big data" for prediction and classification, we next describe how statistical innovations have provided major breakthroughs in the validation of scientific theories based on complex experimental data in biomedical and astrophysics. Because big data typically require variable/hypothesis selection based on some sparsity assumption to make the inference problem feasible, there is contemporaneous awareness of irreproducible research in modern science. In particular, novel statistical methods in post-selection inference and hybrid resampling are presented to address this "reproducible (replication) crisis".
Categorical Data Analysis (ADDED FEE) — Professional Development Continuing Education Course
ASA
Instructor(s): Alan Agresti, University of Florida; Maria Kateri, RWTH Aachen University
This short course surveys the most common methods for analyzing categorical data. The first part of the course focuses on contingency table analysis, logistic regression for binary data, logistic model building, and loglinear models. The second part introduces logistic models for multi-category ordinal and nominal responses and for clustered data using generalized estimating equations (GEE) and random effects. The presentation emphasizes interpretation rather than technical details, with examples including social surveys and randomized clinical trials. Examples show the use of R, with SAS and Stata also for some examples.
Data Science for Statisticians (ADDED FEE) — Professional Development Continuing Education Course
ASA, Biometrics Section
Instructor(s): Rafael Irizarry, Harvard University
Demand for data science education is surging and traditional courses, offered by statistics departments, are not meeting the needs of those seeking training. A popular recommendation for improvement is that computing should play a more prominent role. We agree with this recommendation, but also advocate that the main priority is to bring applications to the forefront. In this short course we will work through some real world data analysis examples and, in the process, introduce skills and concepts not typically taught in traditional courses. Examples include data wrangling, exploratory data analysis, data visualization, reproducible research, and machine learning. We will also introduce tidyverse tools such as dplyr and ggplot2. Throughout the course we focus on statistical thinking and three key skills needed to succeed in data science, which we refer to as creating, connecting, and computing. This course will be of interest to statisticians that want to gain data analysis skills as well as statisticians tasked with teaching a data science courses. Requirements: Understanding of Probability, Inference and R programming. You will need a laptop with the latest R and Rstudio installed and an Internet connection.
Applications of Hot Deck Imputation Methods to Survey Data (ADDED FEE) — Professional Development Continuing Education Course
ASA, Government Statistics Section, Survey Research Methods Section
Instructor(s): Rebecca Andridge, The Ohio State University College of Public Health; Katherine J Thompson, U.S. Census Bureau
This continuing education course will provide an introduction to the use of hot deck imputation with survey data. Hot deck imputation is a commonly used procedure for handling missing data in which each missing value (recipient) is replaced with an observed value from a "similar" unit (donor). Each step of hot deck imputation will be explored in this course, from different ways in which to select a donor unit through methods for obtaining valid variance estimates. Classical hot deck methods will be presented alongside more cutting-edge approaches, including fractional hot deck imputation. All steps will be illustrated with simulated and real data examples from both business and household surveys, highlighting the issues unique to different populations. The course will also present some challenges that arise in the implementation of the hot deck, such as having fewer donors than recipients, and discuss various methods for overcoming these challenges. Attendees will be exposed to both the theoretical and practical sides to hot deck imputation, and examples will be illustrated using both SAS and R. Participants should have some familiarity with survey sampling concepts.
Effective Presentations for Statisticians: Success = (PD)2 (ADDED FEE) — Professional Development Personal Skills Development
ASA
Instructor(s): Jennifer Van Mullekom, Virginia Tech
Public speaking is the number-one fear in America, and yet being able to do so is absolutely critical for success in business settings. Statisticians must be able to effectively convey their ideas to clients, collaborators, and decision-makers.
Presenting in the modern world is even more daunting when speakers have the opportunity to employ slideware, videos, and live demonstrations. Unfortunately, university coursework and professional development programs are often not targeted toward sharpening these skills. This short course, developed and taught by statisticians, will provide an opportunity to learn how to employ different methods and tools in the phases of the Success = (PD)2 framework.
The material covered is geared toward scientific presentations and based on the works of Garr Reynolds and Michael Alley, among others. The course will emphasize the importance of stepping away from the computer to prepare an effective message aimed at your core point guided with a series of questions and tips.
The design phase emphasizes the importance of structure, streamlining, and good graphic design accompanied by a series of checklists. Of course, practice makes perfect, so we cannot skip this step.
Finally, engaging the audience and effectively using the room and equipment is covered in the deliver phase and is complemented with a handy list of dos and don'ts. No matter where you are in your journey for presentation success, improvement is always possible.
We look forward to seeing you in this valuable class, where you can hone your skills! Be prepared for an active class full of discussion and group exercises.
Survey Data Imputation and Analysis Using SAS® (ADDED FEE) — Professional Development Computer Technology Workshop
ASA, SAS
Instructor(s): Pushpal Mukhopadhyay, SAS Institute Inc.
Survey data commonly contain missing values that result from nonresponse. To provide the same complete data to all the analysts, you impute the missing values by replacing them with reasonable nonmissing values. For example, hot-deck imputation replaces the missing values of a nonrespondent unit with the observed values of a respondent unit. Filling in missing values to reduce nonresponse bias is only part of the imputation task. Analyses of the filled-in data should appropriately account for the imputation. We will show you how to create a set of replicate weights that are adjusted for the imputation. Thus, if you use the imputed data along with the replication variance estimation methods in any of the survey analysis procedures in SAS/STAT, you can be confident that inferences account not only for the survey design but also for the imputation. This workshop will show you how to use the SURVEYIMPUTE procedure to perform traditional cell-based hot-deck imputation as well as modern fully efficient fractional imputation (FEFI), fractional hot-deck imputation (FHDI), and approximate Bayesian bootstrap (ABB) imputation. You will also learn how to analyze data sets that contain imputed values by using the survey analysis procedures.
A Thorough Course on Time-to-Event/Survival Analysis: From Descriptive Measures to Complex Modeling (ADDED FEE) — Professional Development Computer Technology Workshop
ASA, JMP
Instructor(s): Ruth Hummel, SAS Institute, JMP Division; Clay Barker, JMP
This course will teach the basics, as well as advanced modeling options and selection techniques, of survival analysis, which is of enormous use in many applied fields. Survival analysis is sometimes called, or is closely related to: Reliability Analysis, Life Distribution Analysis, Life Testing, Degradation Analysis, Time-to-Event Analysis, and Analysis of Censored/Truncated Data. In this course, participants will learn the underlying concepts of survival analysis, what censoring is and how to handle it, the traditional modeling options (nonparametric, semi-parametric, fully parametric with a variety of potential underlying distributions), variable selection techniques, and model comparison. Participants should have a basic understanding of statistics, such as means and distributions. Understanding of likelihoods and generalized linear models could be helpful but is not required.
Empower Your Team and Yourself: Fast-Track to Statistical and Data Science Success Using Minitab's Predictive Modeling Capabilities (ADDED FEE) — Professional Development Computer Technology Workshop
ASA, Minitab
Instructor(s): Cheryl Pammer, Minitab, Inc.; Mikhail Golovnya, Minitab, Inc.; Charles Harrison, Minitab, Inc.
This presentation is directed toward: 1. Applied statisticians wanting to fast-track their ability to incorporate data science and predictive modeling into their repertoire of statistical tools. 2. Educators hoping to augment their traditional statistical coursework with modern machine learning techniques.
As a manager or as a member of a statistical support or educational team, you support statistical analysis needs across your organization. You help colleagues achieve their research goals using world-class, cutting-edge research technology and statistical tools. The Minitab® Suite of Tools empowers the non-statistician domain experts to easily perform hands-on data analytics, including machine learning. If you are the go-to statistician supporting the statistical needs of others, you are likely very busy. Minitab's easy-to-use automated statistical and machine learning software suite makes statistical analysis easy for non-statistician, freeing up your time to focus on the toughest problems. With products like Minitab 18 and Salford Predictive Modeler®, practitioners don't have to be expert data scientists to get the insight they need from their data. Let us show you how to help the professionals and students you work with make appropriate data-driven decisions.
Advanced Methods for Survival Analysis Using SAS® (ADDED FEE) — Professional Development Computer Technology Workshop
ASA, SAS
Instructor(s): Changbin Guo, SAS
Survival analysis provides insights into time-to-event data. Well-established methods focus on right-censored data; these methods include the Kaplan-Meier method, the log-rank test, and the Cox proportional hazards regression model. Survival analysis techniques continue to evolve to meet new challenges, and the latest updates include advanced methods that deal with interval-censored data and competing risks data, as well as statistics for assessing survival models to facilitate risk prediction. This tutorial begins with a review of basic concepts and then presents two sets of model assessment methods-concordance statistics and time-dependent receiver-operating characteristic (ROC) curves. Next, the tutorial introduces analysis of interval-censored data, and how to estimate and compare survival functions with interval-censored data as well as perform proportional hazards regression. The tutorial then turns to analysis of competing risks data, including the use of nonparametric survival analysis and how to investigate the relationship of covariates to cause-specific failures. The cause-specific hazard regression approach will be discussed and compared to Fine and Gray's subdistribution approach. Applications are demonstrated with the survival procedures of recent SAS/STAT software releases.
Formalize the Use of Bayesian Methodology in Your Clinical Trial Framework with NQuery Advanced (ADDED FEE) — Professional Development Computer Technology Workshop
ASA, Statsols
Instructor(s): Ronan Fitzpatrick, Statsols
The cost of failed clinical trials is unacceptably high and the industry needs to focus on ways to reduce the continuously high failure rate. Companies need to make better decisions much sooner in the developmental process. They need to be in a position to recognize which of their early trials have the best potential for ultimate phase III success and then direct much needed resources towards these high-potential trials. Statisticians are uniquely placed to lead their organizations in the exploration of newer or alternative methods that allow for a more insightful decision-making process.
nQuery Advanced now includes a new module called nQuery Bayes, specifically designed for the exploration of Bayesian methodology. This new module includes a suite of sample size tables that allow researchers to formalize the use of Bayesian methodologies into their clinical trial framework, including: 1) Bayesian Posterior Credible Intervals 2) Mixed Bayesian Likelihood 3) Assurance. At this workshop, learn how to: a) Use prior information from pilot studies and other sources to make quicker and better decisions b) Discover the likelihood of a "positive" trial outcome and then make better decisions on which trials show the best potential
Optimizing Your Tuning Parameters to Quickly Achieve the Superior Model Accuracy Expected from Expert Data Scientists and Statisticians Using Minitab's Salford Predictive Modeler (ADDED FEE) — Professional Development Computer Technology Workshop
ASA, Minitab
Instructor(s): Cheryl Pammer, Minitab, Inc.; Mikhail Golovnya, Minitab, Inc.; Charles Harrison, Minitab, Inc.
Advanced modelers must tune their machine learning algorithms to create the best-performing model. For some, "best" means best performance, for others, "best" means the best balance of performance and simplicity, while for others, "best" means something else entirely. Model tuning is often a difficult and time-consuming process involving the generation of many related models. Minitab®'s SPM® automates this process, by automatically creating a series of predictive models using a systematic variation of model parameters. The underlying predictive model algorithm could be CART®, TreeNet®, RandomForests® or MARS®, but the final results are a series of models displayed so that you can easily compare them. This permits rapid perfection of model parameters and serves as a guide to model development while automatically completing these tasks in the same way leading data scientists structure their work. You can use automates to run hundreds or thousands of related models as you look for best performance or the best balance of performance and simplicity. Salford Systems SPM provides 70+ AUTOMATES. A few examples will be discussed in this presentation: Automate Priors and Fraud Detection, Automate Missing Value Indicators and Market Research, Automate Target and Engineering Applications, Automate Sample and Web Advertising, and Automate Shave and Manufacturing.
Generalized Additive Modeling Using SAS® (ADDED FEE) — Professional Development Computer Technology Workshop
ASA, SAS
Instructor(s): Weijie Cai, SAS
Generalized additive models (GAMs) are widely used by statisticians and data scientists when knowledge of the underlying model is limited. For example, GAMs are used in the insurance industry to incorporate geographic location in models for ratemaking. Generalized additive models can be viewed as semiparametric extensions to generalized linear models that relax the assumption of linear effects. You can use GAMs to discover nonlinear dependency structures in your data, which might enable you to develop parsimonious parametric models in addition. This workshop demonstrates the use of SAS/STAT® procedures for fitting GAMs, and it reviews the statistical concepts. Basic familiarity with generalized linear models is assumed.
Bayesian Analysis Using Stata (ADDED FEE) — Professional Development Computer Technology Workshop
ASA, Stata
Instructor(s): Yulia Marchenko, Stata
This workshop covers the use of Stata to perform Bayesian analysis. Bayesian analysis is a statistical paradigm that answers research questions about unknown parameters using probability statements. For example, what is the probability that people in a particular state vote Republican or vote Democrat? What is the probability that a person accused of a crime is guilty? What is the probability that the odds ratio is between 0.3 and 0.5? And many more. Such probabilistic statements are natural to Bayesian analysis because of the underlying assumption that all parameters are random quantities. In Bayesian analysis, a parameter is summarized by an entire distribution of values instead of one fixed value as in classical frequentist analysis. Estimating this distribution, a posterior distribution of a parameter of interest, is at the heart of Bayesian analysis. This workshop will demonstrate the use of Bayesian analysis in various applications and will introduce Stata's suite of commands for conducting Bayesian analysis. No prior knowledge of Stata is required, but basic familiarity with Bayesian analysis will prove useful.
Advanced Methods for Regression and Classification: Extend Your Modeling Toolkit Using Minitab's Suite of Tools (ADDED FEE) — Professional Development Computer Technology Workshop
ASA, Minitab
Instructor(s): Cheryl Pammer, Minitab, Inc.; Mikhail Golovnya, Minitab, Inc.; Charles Harrison, Minitab, Inc.
Linear and logistic regression play a big part in the everyday life of a statistician, but situations are often more complicated. Minitab® has solutions for both! Join us for this presentation, which highlight both classical analyses and modern regression and classification techniques that can take your model to the next level. Learn how to expertly handle missing values, nonlinearities, large datasets, variable interactions and more. Using real-world data sets, we will demonstrate the analysis of messy data with Minitab 18 and SPM Salford Predictive Modeler®. With Minitab technology, you can easily use these state-of-the-art techniques to boost model performance without stumbling over confusing coefficients or problematic p-values!
Propensity Score Analysis and Causal Effect Estimation Using SAS® (ADDED FEE) — Professional Development Computer Technology Workshop
ASA, SAS
Instructor(s): Yiu-Fai Yung, SAS
Applied statisticians and data scientists are increasingly facing data that come from observational studies rather than randomized experiments. Inferring valid causes from observational data is a growing problem for statistical practitioners in applications ranging from health care to marketing to government policy making. This workshop introduces two SAS/STAT® procedures for estimating causal treatment effects from observational data: The CAUSALTRT procedure estimates binary treatment effects by modeling the treatment variable, outcome variable, or both variables; the PSMATCH procedure performs analysis that is based on propensity scores, assesses covariate balance, and creates output data sets that behave like data from randomized experiments. You can then use the output data sets to estimate treatment effects that have valid causal interpretations. This workshop demonstrates the propensity score matching methods, inverse probability of treatment weighting, and doubly robust methods of these two procedures through examples. It emphasizes techniques that promote sound practice and effective communication, such as graphical assessment of covariate balance. It also gives a brief, high-level account of causal inference issues and the principles that underlie the two procedures. Basic familiarity with generalized linear models is assumed.
