Data Visualization thread?

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Cavern of COBOL > Data Visualization thread?

Analytic Engine: May 18, 2009; not the analytical engine

Is there enough interest here for a dedicated Data Visualization thread?
EDIT: Yes and here it is. PM me if you want your message edited or removed.

Stuff that isn't Software

Data Stories is the best (and possibly only) dedicated data vis podcast.

visualising data is an insanely dense collection of project links with interesting commentary.
Check out their yearly reviews:
January to June 2011 July to December 2011
January to June 2012 July to December 2012
January to June 2013 July to December 2013
January to June 2014

Data-vis-jobs is a relatively small Google group that attracts great job/freelance/fellowship opportunities.

The work of designer/statistician Edward Tufte is regularly hyped. I can recommend his first book.

ItBurns posted:

I'll second Tufte's books, not that they need it. I have the one in the OP next to me right now. If nothing else they're beautifully illustrated and you can put them on your coffee table.

MrMoo posted:

I like to think Reuters Graphics has awesome charting for all its TV, magazine and digital work, but I not aware of any of it and certainly working at Reuters we have only access to F/OSS solutions like D3

Both Reuters, Bloomberg, The Times have job openings for data visualisation experts to munge Javasciprt, HTML, CSS for politics coverage, markets, and anything else.

Software

D3.js is a powerful web framework for interactive visualizations using JavaScript/HTML5/CSS.
It covers almost all of jQuery's functionality, though the jQuery thread suggests learning D3 second and warns that D3 has more browser restrictions.
This book is a very approachable D3/web design guide hosted for free. It was so good I bought the Kindle version.
bl.ocksplorer.org is a great search engine for D3 functions that crawls the site where people host their D3 projects.
These two videos give a good intro to D3. The first is from a design perspective. The second goes into detail on the point of D3 and is by a guy that absolutely has to be a Goon.

MrMoo posted:

The spiffiest software for data visualisation is Tableau with hosted and deployed models.

For some lame reason a competitor Datawatch exists, there is a video attempting to explain the advantages. Generally around RAD or "exploratory analytics" which is hilarious as the former CTO announced they don't support that.

Lumpy posted:

Bokeh: http://bokeh.pydata.org is a data vis library that at least one goon works on.

ItBurns posted:

I'm using a couple things not mentioned. The first is Highcharts/Highstock. It seems to be super common as I've noticed it on a ton of websites in lieu of static images of simpler bar/line charts. In some ways it's a less flexible D3, but it's still highly customizable and the built-in types cover a lot of ground with new stuff being added pretty regularly.

http://www.highcharts.com/
Example: http://www.gw2spidy.com/item/46741

Some other options for people who are accustomed to R might be Shiny or ggVis. Neither are very mature, but they're good for simple stuff if you aren't a web designer as they take the guesswork out of making things presentable. ggVis is really similar to ggPlot if you've used that package.

http://shiny.rstudio.com/
http://ggvis.rstudio.com/

Leaflet is a mapping program similar to what you might get from Google maps, or exactly what you'd get on Craigslist (because they use it). Setting it up is really simple, but the actual business of mapping data is probably where it's going to get tricky for some people if you aren't familiar with projections. On the other hand, if you've ever said 'gently caress shape files forever' then this is going to be a breath of fresh air.

http://leafletjs.com/

Open Questions

Ahz posted:

I'm working on integrating data viz into my app and was thinking about the current climate for client-side processing vs. server side.

I can order and group my data as needed for server-side. But I'm wondering if it's worth it to save some processing on my end and put it on the client if most clients these days can handle the calcs averaging datasets of maybe 10,000 rows by 5-10 columns. But them I'm also increasing my transfer, so I just don't know.

Also, are some types of visualizations more/better suited to mobile vs. desktop?

ItBurns posted:

I'm most curious about what people are doing to generate/serve data. Being an R guy I'm doing 99% of my analytics in R and producing JSON/CSV files at regular intervals. It's not real-time and it's a real pain for anything where the user has a lot of freedom.

Plugs

Post your website or portfolio and show off to other Goons

Analytic Engine fucked around with this message at 05:35 on Aug 20, 2014

# ¿ Aug 13, 2014 02:16

Adbot: ADBOT LOVES YOU

# ¿ May 18, 2024 04:37

Analytic Engine: May 18, 2009; not the analytical engine

Ahz posted:

I'd be interested in some good resources.

Edit: I added stuff to the OP.

Analytic Engine fucked around with this message at 07:09 on Aug 14, 2014

# ¿ Aug 14, 2014 07:06

Analytic Engine: May 18, 2009; not the analytical engine

Ahz posted:

I'm working on integrating data viz into my app and was thinking about the current climate for client-side processing vs. server side.

I can order and group my data as needed for server-side. But I'm wondering if it's worth it to save some processing on my end and put it on the client if most clients these days can handle the calcs averaging datasets of maybe 10,000 rows by 5-10 columns. But them I'm also increasing my transfer, so I just don't know.

Also, are some types of visualizations more/better suited to mobile vs. desktop?

These are great questions and I don't know the answers, hopefully more Goons will check out the thread.

I've found interactivity to be harder on mobile since there's no hovering and finger gestures already navigate the browser.
I had a fiddly context-sensitive menu that was unusable on mobile so I added a separate set of big touchable buttons.

# ¿ Aug 14, 2014 23:23

Analytic Engine: May 18, 2009; not the analytical engine

visualising data is an insanely dense collection of project links with interesting commentary.

Check out their yearly reviews of the data vis world:
January to June 2011 July to December 2011
January to June 2012 July to December 2012
January to June 2013 July to December 2013
January to June 2014

# ¿ Aug 20, 2014 05:33

Analytic Engine: May 18, 2009; not the analytical engine

mortarr posted:

I've got nothing public to show, but I just threw this together as a tech demo for management last week. It's an example of call data for water related faults logged over an 18-month period, to show how we might visualise non-financial data. The slider at the bottom changes the selected month, mouse-over changes the info up top.

Tools are pure js + telerik/kendo ui slider and treemap components, which were fairly easy to get going.

MrMoo posted:

I like history of tree maps, never knew they were so complicated. The first problem when looking at a tree map is that the squares for equal values are not always the same size.

If you're interested in tree maps check out Data Stories episode 29 with Ben Shneiderman

# ¿ Aug 22, 2014 00:56

Adbot: ADBOT LOVES YOU

# ¿ May 18, 2024 04:37

Analytic Engine: May 18, 2009; not the analytical engine

There's a new book on Data Viz by Tamara Munzner, possibly the first comprehensive textbook covering every aspect. Harvard's CS department is using it and industry professionals are excited.

http://www.amazon.com/Visualization-Analysis-Design-Peters-Series/dp/1466508914

Analytic Engine fucked around with this message at 00:46 on Jan 26, 2015

# ¿ Jan 24, 2015 19:55

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > The Cavern of COBOL > Data Visualization thread?