How to better showcase your research results

How to better showcase your research results

"Cos Editorial Press" The writer is a member of the American Statistical Association (ASA) and Monash University professor Dianne Cook. Her research interests include data visualization, exploratory data analysis, multivariate approaches, data mining and statistical computing. has been involved in the production of software Xgobi,ggobi,cranvas and several R packages. The original text is published in the R journal, this article by Chen Yan Translation, Ko Tao, Xiaonan and Yihui Xie Review, Wang Xiaoning edit.

Summary  

With the International R User Conference "user!2011" approaching, many attendees may be thinking about how to focus their ideas through a presentation, and this article provides some advice on issues such as speaking and making posters.

Background

During the upcoming college job interview, I'm going to introduce my PhD research project, and as soon as I finished the interview exercise, my mentor, Andre Buga, asked me to sit down: buja my speech! I started off with a lot of speakers at Rutgers University's weekly seminar, showing the details of my research work through a slideshow, but Andre said that might be a good fit for the presentation, but not as the best choice for the lecture. We listed the key issues in my research and then inserted a slide to simply write, "Listen to tell for your funeral." We explained the methodology in the next few slides and gave the answer to the question at the close of the report.

This "want to know funeral how, please listen to tell" deeply touched me.

Core points

Schoeberl&toon (2011) Websites and Marle (2007) websites have some very useful suggestions for making scientific reports. Here are some of the points that I've summed up from my own experience:

• Make an overall plan, carefully arranging the opening (including how to inspire the audience's interest), middle and end sections. Make a slideshow frame outlining the issues you want to tell on each slide.
• Avoid a single slide list (ppt syndrome), or a page after page of equations.
• Build your theory. For a traditional listener, you may need to write some equations or a core part of the proof. For those who care about statistical calculations and computational statistics, you may want to show a piece of code-such as a piece of writing that is very elegant, or that solves a critical problem. (My colleagues and I occasionally joked that it might be more interesting to see programmers writing programs than listening to reports.) This is, of course, half-joking, but sometimes it's instructive to look at a seasoned Programmer's debug program. ）
• Use a drawing based on data. But do not say "we can see from the picture ...", to add some explanatory statements, such as "because these points conform to a certain curve form." In other words, to explain the characteristics of the graphic, so that people can keep the thinking continuity.
• Use graphics or cartoons to help explain some of the concepts.
• Use pictures to create visual stimuli. But this is not because the picture "is a tool for ignorant people to know common sense", nor is it because pictures can "prevent some dumb listeners from dozing" (tufte,1990), but because the graphics are beautiful and inspire the audience.
• Tell a story.

Practical advice

Many statistical beamer templates for home latex make slides. It's handy for editing equations, and it's beautifully formatted. And it has great flexibility in the making of text colors, font selection, navigation bars, and graphics and animations. But I prefer the keynote of the Apple Computer. It's much more flexible in formatting, and it's able to use very personal fonts (such as my handwriting), and it can seamlessly combine animations and movies. There is a "Tmp.tex" file on my computer that is designed to collect equations with all the equations I've used, and I cut and paste the past from the preview file in the PDF. When adjusting the size of equations and graphs, keynote can guarantee the image quality of the original, which PowerPoint cannot. Keynote and Tex, you can store graphics in separate files, "Slide.key" may look like a folder, but it is actually a directory.

Robbins (2006) is a basic guide to drawing. The Ggplot2 package (wickham,2009) in R provides a beautiful, well-pivoted graphical default color.

The power of the poster

• Pay attention to the layout and image coherence of the poster.
• Identify the major issues your report wants to articulate and identify who your audience is.
• Choose your color scheme, and consider whether it is easy for color-blind people to identify, to avoid the combination of special colors, such as: red, yellow, black, they together is the color of the German flag.
• Select the font size and font. The topic wants the font size about 100pt, the small title font size 50pt, the body size at least 25pt. To avoid all caps.
• The data-based graph has a good focus. A graph of contact context can help people quickly grasp the focus of the data, but also allow people to see something familiar and attract their attention. High-quality drawings, such as those made with R, are important.
• Animated audio can attract the attention of passers-by, but they cannot be used to attract listeners to participate in group discussions.

You know, the statistical conference posters that do not work well abound. Therefore, it is not appropriate to Caishuxueqian for yourself by "everyone else is doing the poster." With the gradual deepening of our understanding of good poster planning, the standard for each poster is becoming more and more high. We can find very good advice on making posters on the Cape Higher Education Alliance (2011) website, and the Purrington (2011) website has a valuable discussion on designing scientific posters. The Data Expo Competition (ASA,2011), which is combined with the Joint Statistical Conference, often has some very good posters to make. In addition, we can also find posters of previous international R user conferences on https://www.r-project.org/conferences.html.

A responsible listener?

I occasionally or more often find listeners praising the highlights of a particular report, but it's clear they don't even know what the report says. It is therefore incumbent on the audience not to show praise because their good impressions of the speaker are easily magnified by the speaker. The audience also has the right to expect the speaker to make the report clear and easy to understand, and to explain the report clearly.

Conclusion

It is important to remember that it is an honor to report in front of a colleague-not everyone has the opportunity to speak their mind and be heard, especially in high-spec meetings (e.g. user! ）。

A lot of people are interested in TED Talk (rosling,2006), and I've recently been intrigued by a report by Chris Wild (2009), which is an excellent statistical report.

Reference books

[1] American Statistical Association. () Data Expo posters URL http://stat-computing.org/dataexpo/

[2] Beamer developers. Beamer-a LaTeX class for producing presentations. URL Https://bitbucket.org/rivanvx/beamer/wiki/Home.

[3] Radford M Neal, Bayesian Learning for Neural Networks,, 1994

[4] Cape higher education Consortium. (information) Literacy URL http://www.lib.uct.ac.za/infolit/poster.htm.

[5] D.cook. (improving) statistical posters. Urlhttp://www.amstat.org/meetings/jsm/2008/pdfs/improvingstatisticalposters.pdf.

[6] B.dougherty and A.wade. (+) URL http://www.vischeck.com/vischeck/.

[7] Etre Limited. (+) URL http://www.etre.com/tools/colourblindsimulator/.

[8] R.ihaka, p. Murrell, K. Hornik, A. Zeleis. (+) Colorspace:color Space manipulation. URL http://cran.r-project.org

[9] E.neuwirth. () Rcolorbrewer:colorbrewer palettes. URL http://cran.r-project.org.

[Ten] C.purrington. (Advice) on designing scientific posters. URL http://www.swarthmore.edu/NatSci/cpurrin1/posteradvice.htm.

[One] n.robbins. (2006) Creating more effective Graphs. URL http://www.wiley.com.

[M.schoeberl] and B. Toon. Ten Secrets to Giving a good the scientific talk. URL http://www.cgd.ucar.edu/cms/agu/scientific_talk.html.

[E.tufte]. (1990) The Visual Display of Quantitativeinformation. Graphics Press. Cheshire, CT.

[A.J] van Marle. (+) The Art of scientific presentations. URL https://www.cfa.harvard.edu/~scranmer/vanmarle_talks.html#Technical_preparation.

[]15] H.wickham. Ggplot2:elegant Graphics for data analysis. UseR. Springer.

[+] C.wild.  Early statistical inferences:the Eyes has It. URL http://www.stat.auckland.ac.nz/~wild/09.wild.USCOTS.html.

How to better showcase your research results