Big Data Visualization caries with it different requirements than the similar business data visualization requirements that you may find in traditional business intelligence solutions.
With Big Data Analytics, you are likely going to need to provide visualization capabilities to more than the general knowledge worker community that would typically have requirements for no further detailed data than the aggregated business-level view. In order to support ad-hoc data discovery and for functionality needed by your data scientist community, you will need to provide data visualizations that can help to provide context and meaning behind very large data sets with possibly millions of individual data points.
When I am analyzing large clicksteam or Web Analytics data sets, I like to present the data in a diagram like a SanKey (I got this from OUseful.info):
This is a very helpful way to demonstrate data relationships, paths, flows and which path or input has the most impact on the output.
Common tools like SSAS in SQL Server (the diagram below is from a bidn.com tutorial) can show diagrams in the data mining tools from Visual Studio (or Excel, for smaller data sets) that demonstrate classification, relationships, paths, etc. to the analyst.
That is a Microsoft tool only available in Visual Studio or Excel and very useful. But will not always scale to the Big Data requirements that your larger projects may have that include sensor data, clickstream, etc.
But in traditional BI tools, there are a variety of visualizations that work well for both dashboards as well as ad-hoc data discovery analysis, which is aligned with the Big Data / Data Scientist audience. If your data scientists are going to leverage Big Data tools to access deep granular data in Hive / Hadoop, then the number of data points that you’ll have to graph will not be possible as a traditional time series or X-Y graph.
Tableau 8, for example, now includes heat maps, which is one my personal favorite tools to take big volume data and aggregate those into an easy-to-read format in a chart: