5. I have used tableau in my professional activities and it is indeed a great tool but most of the times it is required to understand & filter the data before actually using it. The issue of outlier-error that you have pointed out applies here as well.
Thanks everyone! I hope this was helpful. Just remember your audience comes first!
Feel free to check out my site and email me if you have further questions or if you'd like me to come to your place to help talk more about data visualization and presentation techniques.
@jchang - that sounds very logical and is an idea I can definitely take to my collagues (I get to "book report" this little chat too lol)! I think I will be adding a web based visual API or linking capabilities to our requirements for our systems selection.
@Jonathan - Thanks for the information - it was very uncluttered :)
As an example of data visualization via map format, here in Austin we're debating alternative corridors for urban rail. The following map book has various data visualizations of demographic onfo such as population, emplpoyment, pocert, auto ownership, etc. which is associated with the different candidate corridors:
However, many of the projection in future years are being criticized, since projections are a function on assumptions (and the intentions of those who do the projecting). It's a GIGO problem...
@Akyut - I'd tend to go for an in house web server accessed via single sign-in to your network. it's relatively easy to set up, free (apache, anyone?) aside from a machine to run it and a resource to monitor it, and completely internal to the company.
@jchang - that is a good point. We have a engineering department that could, in theory, build out some visualization options. Based on your experience - focusing on the web-based publishing paradigms would be the way to go? Use the different tools and then tap into displaying them via a web-based application - even if that application is inhouse?
@ASykut In my experience, a lot of people still actually like paper! But I also find that many people can fairly easily manipulate interactive Excel dashboards (with only a little coaching). Tableau visualizations are also fairly easy to use. The downside is that if you are using the public version of Tableau, the information is posted online, so you might not be able to use it for internal visualizations.
when you are doing distribution (vs. accessing a common URL/sharepoint), you need some sort of enterprise push solution. That's the one part where the free tools tend to fail because they're just not necessarily designed for that type of distro. they tend to focus more on web-based publishing paradigms (which are more efficient, quite frankly).
@Jonathon - I will add those to the list. Based on the need to use multiple tools, and the fact that clients are often looking for a weekly and daily dashboard showing this information - what are the more common distribution methods? I know, for example, that Sharepoint supports PowerView. What are common options you all have run into?
@MFLoGrasso - have you used them on PowerView? It did it all intuitively with as you said, the same data source in the data model. I also like in power view how you can simply click on a category line for example and it will filter all of the charts in the same view to that line.
== sounds like you're introducing bias with color choice. How concerned are you about that?==
Keep in mind that a lot of work in public transportation and urban planning involves persuading the audience -- often, political leaders. Data visualization is an important tool. Need I say more?
@Jonathan - I had low expectations and I think they more then delivered. The only low point I have on it is their map. I love how easily I can map data, but it is very limited on how it maps data. I am investigating an addin of PowerMap. Heat maps and point maps are highly desired by our clientelle.
@Jonathan - Awesome. I just tested out Excel Office 2013 PowerPivot and Powerview capabilities this past week as tools to present our analytics and BI data. I found it works very well compared to some third party dashboards I have reviewed. We are researching "dashboard" solutions and will add those you mentioned to our list!
== There's really no sense using multiple colors in a simple column chart, especially because some colors (esp red) can be perceived as highlighting those series versus others. ==
I gotta admit I use red often to either indentify the "bad guy" in the data set, or to otherwise call attention to something I want to focus on...
@InsightsRUs There is one great way to use pie charts and that is to remember that the "part-to-whole" comparison is just that--compare single parts to the whole. So, next time, instead of a single pie chart with, say 7, slices, try 7 different pie charts with the one slice of interest in some color and the rest of the pie in a single (less pronounced) color.
@Lyndon_Henry Yes, I prefer to use color as a way to help encode--or emphasize--particular points. There's really no sense using multiple colors in a simple column chart, especially because some colors (esp red) can be perceived as highlighting those series versus others.
I think a breakthrough came recently, though, when I had been providing simple graph after simple graph after simple graph, and there finally came a request for some rather complex relationships to be summarized. Since I leave my graphs pretty uncluttered in the first place, the addition of additional info on-screen was easier for my audience to navigate.
As an aside, it ended up being an Excel scatterplot with error bars, a line for the company-wide mean, and line's for segment means.
@ASykut That's a great question. I think you can do a lot with Excel--which most everyone has--that most people don't realize (see some of my blog posts on AllAnalytics to see what I mean). Tableau is also a tool that can be used for both. Similarly, the programming language R is open source (ie, free) and can be used to create great graphics. The program Inkscape is basically the open source (free) version of Adobe Illustrator.
@Lyndon_Henry But the problem with adding labels is that it starts to clutter the graphic (especially if it's a slide in a presentation). More generally, an important question to ask is, "What is the purpose of this visualization?" You're having to add data labels to a 3D graph because the graph itself is an ineffective way to communicate information. Going to 2D creates a more effective graphic and thus promotes better understanding.
Seems to me data viz is most useful is showing relationships among different data groups and revealing data patterns. I do use color a lot because I think it makes the presentation more visually appealing, whereas monocolor would seem boring to me. But you seem to prefer monocolor unless there's a specific data-related reason for the color?
@Jonathan - in your presentation you discussed audience. The difference between analyse and presentation. Do you find many tools that are affordable (I primarily work for non-profit organizations) that are strong on BOTH analysis and presentation?
@Jonathan - thanks. I think you are right - your simple graphics in this training were very compelling. It may be a good counter point to include it in demos about say...dashboards. a Why we product THIS (uncluttered) not THIS (cluttered). THis has been one of our biggest sticking points!
@jchang Well, the best place to get training is my website! :)
There are a few others out there conducting trainings, but they are far and few between: see visualisingdata.com; storytellingwithdata.com; and http://stephanieevergreen.com/; and perceptualedge.com
@ASykut I think you can sell the charts by demonstrating that the 2D representation more accurately shows the data than the 3D. If that requires a quick training--for example, a single slide with both versions--then that might be worthwhile.
@ASykut, interesting question. Jonathan, what do you think? @ASykut earlier asked: Question: How would you phrase it to a client who thinks 3D graphics and picharts are neccesary. How do you sell the simple charts? Do you give them a mini training session during your sales pitch?
I think it is important to mix it up when you have a big presentation. No one wants to look at slide after slide of line charts or pie charts. It's good to mix it up and make it interesting to the viwers. I'm not a huge defender of pie charts but what I am trying to say is that I have seen them used very successfully in sales and marketing groups to show a non-complex situation in a quick, visual easy to remember way.
@Beth - that's a good question. My company has TONS of analysts but very few of them are doing "analysiis" - most are doing reporting. And even the ones doing "analysis" aren't doing visual analysis or worrying about how to properly present their reports. We need more training but the value of this kind of thing is not necessarily seen immediately.
It depends on what type of data you're looking at. Pie charts are very efficient, easy to grasp, quick visualizations for market share for example in markets where there are only a few players and their share is not similar. If you are trying to show more complicated segmentation with small differences or ranking, of course you would use a different visualization. It all depends on your purpose aa the presenter explained.
@jchang, that's the issue I face. My direct supervisor part of the company's Senior Management team, so they rarely want to have to take too many additional steps themselves unless those additional steps also contain actionable information
Hello everybody. I'm looking forward to today's lecture with Jonathan Schwabish, a passionate create of data visualizations. If you haven't already downloaded his slide deck, please do so now from the link above. This slide deck is newly posted as of a couple of hours ago. So, if you downloaded earlier, please get this new version. Our previous lecturer, Hyoun Park, had to drop out for personal reasons and Jonathan has gratiously jumped in today!
A2: ALL ANALYTICS ACADEMYVIEW CLASSES +
Get up to speed with emerging analytics technologies including Natural Language Processing, Edge Analytics, Machine Learning, Real-time Analytics, and Augmented Analytics. These expert-led sessions are for analytics leaders, professionals and business users.