How Good People Make Bad Graphs: Unintentional Fibbing with Visualization
Reasonable goals can often result in incredibly misleading visualizations. This talk will show several examples illustrating how easy (and how dangerous) it is to lie with data visualization, and discuss some of the goals that may have led to that result. Then, we will explore how to rework those visualizations to still achieve those goals without distorting our perception of the data.
I will post slides here after the meetup.
I will hold bonus office hours this week on Tuesday 05/06 and Thursday 05/08 after class in the 5th floor lobby to help with any remaining bugs you have in your final projects. Also, please post on Piazza for help instead of sending me email. (This actually has a better chance of getting a response from me, since my email is flooded.)
The IEEE Conference on Visual Analytics Science and Technology (IEEE VAST) hosts a yearly visualization challenge. The challenges are open to participation by individuals and teams in industry, government, and academia.
All participants are invited to submit a two-page summary of their entry for inclusion in the IEEE VIS 2014 electronic conference proceedings, and attend the VAST Challenge 2014 Workshop, to be held in conjunction with IEEE VIS 2014 in Paris, France.
There are 3 mini challenges and 1 grand challenge. The data involved includes text (including streaming), network, geospatial, temporal, and transaction data. Visit http://vacommunity.org/VAST+Challenge+2014 for more details.
The submission deadline is July 8, 2014 at 11:59pm Pacific Daylight Time (UTC/GMT -9 hours).
You can find details about your final presentation, including the randomly generated presentation order, on the Instructure Canvas website.
Everyone will have 7 minutes total to present, including setup and questions. You will need to be prepared and practiced to give a strong short presentation!
The Data Visualization Speaker Series presents:
Monday, April 28, 2014
6:30pm to 9:00pm
McLaren, Room 250
Reza's talk will explore the process of designing and implementing interfaces and generative visualizations. He will showcase his most recent works and address the aesthetic considerations and challenges involved in creating each piece.
|6:30 – 7:15||Doors open, socializing, food, drinks|
|7:15 – 8:15||Speaker's presentation|
|8:15 – 9:00||More socializing, Q&A|
Bio: Reza is a computational designer and creative engineer. He uses code to express himself, and creates tools and libraries to help others create. He is the author of ofxUI, one of the most popular add-ons for openFrameworks (an arts-engineering toolkit). Though his works is highly visual, he is interested in the ideas and implications of his tools, works, and methodology. As an artist and engineer he is deeply interested in the feedback cycle that occurs when customs tools are written to accomplish an artwork, and how those tools can enable artworks to go beyond what the artist had originally envisioned. Reza is excited to find those happy accidents, beautiful moments of interaction and magic that comes with making something novel with technology. He hopes to change the world by making a difference in how people use technology, design products and experience new media art and entertainment.
Notice: Attendance at this event is mandatory, and will replace lecture on Tuesday 4/29. (We will not meet on Tuesday 4/29 as a result.)
I just want to say that, overall, your homework feedback is fantastic. Thank you for providing useful, detailed, and positive feedback to your classmates!
Submit your redesign from today's exercise at:
You can download a version of the original data (plus some extra context) at the following link:
The following links are where you can find the original visualizations:
Here are some related links that discuss the original visualizations:
There is also quite a bit of additional discussion on Twitter.
We will start using Piazza for all homework questions in this class. This will provide a forum for people to see the responses to common questions, and share various tips and hints when it comes to wrangling
I will start redirecting any questions sent via email to Piazza instead. You can post anonymously if you feel uncomfortable attaching your name to a question, or privately if you feel uncomfortable sharing your code. (Please note the instructor can always see who posts on Piazza, but other students will not see your name if you post anonymously.)
I released the grades for homework 2 in Canvas and unlocked the assignment so that you can view the rubric used. I really enjoyed seeing the creativity of everyone with the customizations. There were some really neat visualizations!
To select which students would have homework 2 graded, I randomized the list of all students using https://www.random.org/lists/ and selected the top half of the list. If you received a 0 on this assignment, this means you will have homework 3 graded instead. Note that Canvas is now setup to drop the lowest grade in the homework category.
I will repeat this process to determine who will have homework 4 versus homework 5 graded, but will not release the randomization until after homework 5 has been submitted.
The Schedule has been updated with links to the demos and homework for text visualization.
Important We will hold class from 2:00pm to 4:00pm on Thursday, April 17, 2014 to accommodate for the Holy Thursday university holiday.
Check out http://app.raw.densitydesign.org/ for an elegant way to create web-based visualizations.
When asking for help, please include (1) the error message you are receiving, and (2) a link to your code in your GitHub repository.
Here are some upcoming assignment deadlines:
Tue Apr 08 11:59pm – Homework 2 Feedback
Thu Apr 10 11:59pm – Homework 3 Multivariate
Tue Apr 15 11:59pm – Homework 3 Feedback
Tue Apr 15 11:59pm – Project Dataset
You will get time this Tuesday in-class to work on the homework feedback. After that, you are expected to complete the feedback by the end of the day every Tuesday.
I have extended the deadline for Homework 2: Interactivity by one day. It is now due Friday April 4 at midnight.
If you are having trouble keeping colors consistent in your plot when you filter out data, please see the
ConsistentColor.r demo on GitHub.
FYI: It looks like the
shiny package has a new and improved website at http://shiny.rstudio.com/ with much more documentation than before.
You can now see your homework 1 grades (and comments) in Canvas. I was lenient grading this homework. This goal of this homework was to get your GitHub repositories setup correctly and get you started in
ggplot2. This will not be the case for future homework!
Please keep in mind the following for your future submissions:
Please do not have
setwd() in your R scripts. This will prevent me from being able to run your code directly from within R. You can use Projects in RStudio to manage your working directories instead.
If you do not have enough to write about in your discussion, then you have not customized your visualization enough. Please include what you changed and why you made those changes. You can also discuss what you tried that did not work, and what you learned about the dataset from your visualization.
Please place your discussion in
README.md since this is the file that automatically displays when I visit your repository. Also, consider using headings in README.md to separate your discussion. This will make it easy to match your discussion to its associated plot.
Overall, I am thrilled to see so many of you trying out different things in
ggplot2. I look forward to your future submissions.
Now, for the bad news. I may not always be able to look at everyone's homework submission directly. In that case, I will randomly select students to grade. Everyone else will receive a 100%. Students that do not have a homework graded are guaranteed to have the next one graded. If you did not get your homework graded and want feedback, feel free to ask me for feedback after class on Thursdays.
Same numbers... different stories:
This is a good example of why bar charts should usually start at zero. Strangely enough these charts come from the same organization! See mediamatters.org for more information.
An Advanced Homework Workflow guide has been posted using SourceTree. This guide is only for advanced users unafraid of revision control, and not necessary to submit homework in this class. In particular, it allows you to update your homework repository anytime I update a homework description in the master repository.
Those of you that have filled out the GitHub Survey on Canvas have been added to the MSAN 622 Organization on GitHub.
Your membership to this organization is private by default. You may change this to public by logging into GitHub, visiting https://github.com/orgs/msan622/members, and clicking the "make public" link next to your account. If you want to see what is visible to the general public, log out of GitHub and then visit this page.
There will occasionally be in-class exercises. If possible, please bring a laptop to class. You will need to have R installed.
Some students have expressed interest in learning D3 instead of R and ggplot2. Since we have a mix of backgrounds this semester, we will stick to R and ggplot2 as the primary tool.
If you want to self-learn D3, you will be able to submit homework and projects using D3 instead. No instruction will be given in D3, however.
Welcome to the Spring 2014 course website for MSAN 622. I will be migrating to using GitHub Pages and Canvas for this course. Please be patient as the content is moved over!