InetSoft on DM Radio: Appropriately Visualizing Big Data

This is the continuation of the transcript of DM Radio’s program titled “The Eyes Have It: Ten Reasons Why Data Visualization Rocks.”

Eric Kavanagh:  Yeah, that’s a good point.  What are some of the, to get to that other question, what are some of the bigger mistakes you have seen people make when trying to use data visualization to communicate something?

Suzanne Hoffman:  You know what, mistakes not so much as just poor behavior as they constantly use the same type of visualization for data that would bode well in a different one.  Again, this whole concept of a guided analysis or being able to understand what best practices, or being able to hand best practices.  

And I think Byron actually said it earlier, do I need a cross tab for this as opposed to having a bar chart or having a pie chart or having some other visualization, which is more appropriate to the data in hand.

So it won't be quite as eye catching, and it won't be as impactful if I use even the wrong data element in the wrong visualization. I won't be able to get to the underlying information that I need as quickly, nor will I be able to blend it as quickly. 

Look at a Huge Array of Information

If I am really going to use a scatter plot, that enables me to use and to look at a huge array of information, across product lines or across timeframes whereas if I am just using a map or if I am using a bar chart, I don’t have that same interaction in contextual fields with the data that I need.  So I think that that’s actually a big issue, and that’s for people who already have embraced visualization.

Eric Kavanagh:  Yeah.  I guess maybe Mark, I will bring you back in because you have been studying a lot about Big Data, and of course one of the comments I heard Robin Bloor make one time which I thought was rather astute is that it’s just not unstructured data. It's just largely un-modeled data because of course a lot of that so-called unstructured data is highly structured. 

You made that point in a webcast just yesterday I believe.  But it seems to me that these challenges of trying to find the best data visualization approach or tactic to use in a particular case in the world of highly structured data like transactions and who bought how many products where and so forth gets a whole lot more complicated when you start working with these different kinds of Big Data, right?

Mark Madsen:  Yeah.  I think when the data is more variable, and there is more variety of it. You think back to the early days of BI data warehousing with a couple of fact tables and a handful of dimensions.  And now you have got hundreds of tables, and you are throwing in data with a huge number of attributes that you don’t have, as Suzanne said guardrails around things anymore.  

And with this huge variety, you also can’t model everything in very effectively and so the tools need to pick up some of the slack and give you the ability to look at it five different ways.  And to do that in a spreadsheet or BI tool certainly isn’t easy, and so it reduces the effectiveness of both the tools and the information.

Read the top 10 reasons for selecting InetSoft as your BI partner.