Statistical Graphics Working Group Seminars

  • 12/10/2014 Adam Loy Better Diagnostics for Linear Mixed-Effects Models Using Visual Inference There are numerous situations in which conventional diagnostic tools for linear mixed-effects models perform poorly in practice. For example, the empirical distribution of the predicted random effects does not resemble the theoretical distribution without strong assumptions, rendering traditional use of Q-Q plots and tests of the empirical distribution function ineffective. In this talk we will discuss how visual inference can be used in such problematic situations, and how it provides a unified framework for model selection and diagnosis. In the second part of the working group I will talk about my experience teaching at Lawrence University, a small, private, residential liberal arts college. This discussion won't be formal, so bring any questions you might have!
  • 12/3/2014 Graphics Qualifying Exam
  • 11/19/2014 Carson Sievert Web scraping in R A staggering amount of information is freely available on the world wide web. Acquiring and transforming desired information into a usable form is often a difficult task. This talk will cover a common task of transforming HTML/XML/JSON files into data tables using R.
  • 11/12/2014 Gray Calhoun, Asst Prof in Econ A visual analysis of some U.S. macroeconomic series We will look at some of the important US macroeconomic time series, GDP, unemployment, and inflation, and will discuss aspects of their evolution over time. The talk will focus on broad patterns of the series' behavior, not specific modeling details, and will cover both long-run trends as well as business cycle effects.
  • 11/05/2014 Eric Hare and Andee Kaplan Data analysis with friends (Part 2). In this talk we will perform a group data analysis real-time in the ISU graphics working group using CoLaboratory (see here for installation if you couldn't make part 1). We will explore some of the economic benefits of choosing a particular graduate major based on data from the American Community Survey 2010-2012 Public Use Microdata Series.
  • 10/27/2014 Hadley Wickham, RStudio Pipelines for data analysis Over the last year and half, three things have had a profound impact on how I develop tools for data analysis: Rcpp, writing the advanced R book (http://adv-r.had.co.nz/) and the pipe operator (%>%, from magrittr). In this talk, I'll focus on the pipe operator and how it's influenced the development of tidyr, dplyr and ggvis, the next generation of reshape2, plyr and ggplot2. Come along to learn about why I think pipelines are awesome and see how pipelines + tidyr, dplyr, and ggvis can make your data analysis fast, fluent and fun.
  • 10/22/2014 Eun-kyung Lee, Visiting associate professor from EWHA Visual model validation in pharmacometrics Phase I clinical trial is the very important step in the new drug development procedure. The population pharmacokinetic/pharmacodynamic model is used to figure out the kinetic properties of new drug compound in phase I clinical trial. The ultimate goal is to select an appropriate dosing regimen, optimize further clinical trial design, and make informed go/no-go decisions. This modelling procedure needs to be carefully validated. In this talk, I'll introduce various graphical model validation methods in the pharmacometrics.
  • 10/15/2014 Eric Hare and Andee Kaplan Data analysis with friends using Digital Oceans This is part one of a two part seminar for the ISU graphics working group. In this talk we will discuss a new tool for collaborative data analysis developed by Project Jupyter - CoLaboratory. We will give some background information on the tool, and then spend the majority of the session walking through installation with the group. In Part 2 (Oct. 29) we will perform a group data analysis real-time in the ISU graphics working group using CoLaboratory.
  • 10/8/2014 Xiaoyue Cheng Visualize areal data via cartograms Cartograms provide a way to show geographic data where areas match the value they portray, but the geography is still recognizable. I will discuss implementing the algorithms into R.
  • 10/1/2014 Heike Hofmann, Lendie Follett and Adam Loy Q-Q plots - the power of our eyes A description of an experiment exploring how people read Q-Q plots, should they be rotated to be viewed in comparison to a horizontal line, and how they compare with classical tests.
  • 9/24/2014 Sam Tyner My summer internship with Kingland Systems How I analyzed data while preparing for the PhD prelim.
  • 9/17/2014 Di Cook, with Luke Fostveldt, Alex Shum and Ian Lyttle Seven Things That Affect Education Around the World The talk describes the visualizations produced for the OECD data challenge at useR! 2014.
  • 9/10/2014 Ian Lyttle, Alex Shum Summertime with ggvis and shiny A summer internship at Schneider-Electric in Cedar Rapids, Iowa.
  • 9/3/2014 Carson Sievert Producing interactive animations with animint A google summer of code project work mentored by Toby Hocking.