We will talk informally about things graphics we did this summer and during the Joint Statistical Meetings in Vancouver.
We will also discuss what we want to do for the remainder of the semester. There will be cookies! Everybody is welcome!
Read more →
Cleveland and McGill’s examination of the accuracy of perception of different graphical objects underlies the conventional wisdom that guides our selection of plot types and choices. If you’ve ever wondered why, exactly, pie charts are bad, this paper is the place to start! The paper breaks down graphical perception into a series of elementary tasks which we use while reading charts. We’ll discuss the paper and its implications for creating the most effective plots.
Read more →
Visual inference in RNA-sequencing data In a visual lineup, actual data is plotted among null data, and humans are asked to judge the plot that is different. Creating visual lineups of datasets allows us to assess statistical significance of visual findings, while connecting exploratory data analysis with inferential statistics. We will discuss some of these concepts. I will also present visual lineups for soybean and honeybee RNA-sequencing datasets. This is a work in progress, and so I would be happy to hear any of your advice and input.
Read more →
Using R to create browser-rendered interactive visualizations has long been an interest for Graphics Group, and for the larger R community. One of the most-promising efforts out there is the Vega-Lite library: https://vega.github.io/vega-lite/. Bob Rudis is coordinating an effort to provide a native R-interface to the Vega-Lite library: https://github.com/hrbrmstr/vegalite.
In the Python world, Jake Vanderplas and coworkers offer an API to the latest version of Vega-Lite, the Altair library: https://altair-viz.github.io. In late March, RStudio released to CRAN a package called reticulate: https://github.
Read more →
Parallel and Distributed computing is very helpful for Big Data analysis and cloud computing. Besides, for bootstrap, simulation and machine learning model estimation, this technique is extremely helpful to reduce computation time. In this talk, I will briefly discuss about how parallel and distributed computing works and then demonstrate some R packages (‘parallel’, ‘doRedis’. ‘doParallel’) that can be used to parallelize computation.
Read more →
