Experience in using Visualization Tools for Analytics

I have been fortunate in having experienced multiple data visualization tools for data discovery and interpretation in building analytical products & solutions. I am sharing summary feedback on some of these, which I hope the community can benefit from.

Visualization is a key medium for communicating analytical insights, so getting the right tool for the right use case is critical for overall success of an analytical engagement or product.

1. Tableau

An excellent desktop visualization tool with a rich library of visualization widgets. Users can "self-help" themselves to create strong visualization based analytical stories. Lot of the work is click-and-drop type, but you need to get familiar with using the ToolTip feature and also not been constrained by what Tableau is recommending you (use your imagination fully). Having a geo-mapping tool in-built was a definite plus. Finally, its well priced!

On the flip side, the tool doesn't seem to work well with complicated schema's of database (I didn't try the hadoop connector yet) and it expects you to primarily feed a flat file to it (It does have the ability to join tables etc., but not sure how robust it is for complicated structures). The layout manager was somewhat basic and we struggled to finely control final layouts. Also, since it runs in-memory, loading times initially can be an issue, also you need to invest in good hardware to get performance.

Overall a great analyst and desktop visualization tool, but I am not sure this is the choice for more industrial business analytics solution.

2. QlikView

I really enjoyed working with this data visualization tool. It has all the standard visualization widgets and the macro language was very powerful for data modelling and creating cubes and performance was excellent (it's another in-memory BI tool). What I really liked was the ability to define and build a full analytical application using QlikView with ability to download processed data as appropriate. We were able to rapidly enhance the solution based on client feedback without much heartburn and its data filtering capabilities are strong.

On the flip side, the tool is not cheap and a 25+ user license with good performance hardware is in the ~$100K range. You definitely need a BI Analyst with strong SQL and data modelling skills to build applications on QlikView, this is not your desktop quick-and-dirty type of application (a la Tableau.). When I used this, I didn't see a native geo-mapping solution in-built, there were third party extensions available.

Overall, great solution to build industrial analytical application with strong performance, but watch out for the budget!

3. Jaspersoft & Highcharts

I love the open-source nature of this toolkit(s). We are currently building a sophisticated analytical application with lots of transaction data and strong analytical content on it. But this solution is not for the faint-hearted. Integration with Highcharts was somewhat cludgy (the native Jaspersoft supports Highcharts 2.x and not the latest 4.x). The data modelling capabilities are very good in this tool. The dashboard manger is sophisticated (including iReports) and you can fine tune the layout in a granular fashion. The Highcharts library is excellent (I can’t stop praising it) and these guys keep adding new visualizations to the toolkit on a frequent basis. The new geo-mapping solution called Highmaps is a great example of them responding quickly to user community feedback. The quality of HTML5 outputs is excellent. Can’t beat open-source for price!

On the flip side, you definitely need a good BI Analyst to take advantage of the features of this combined solution. Its integrates well into the overall stack of Talend ETL, connectors to all databases and type (relational and non-relational) and its built ground up on Java which makes it very extensible. Working with Highcharts 4.x required strong technical work, but it was well worth the effort in the end. You need to be patient to build a quality application using this (not as nifty as Qlikview) but results you will get.

This list is definitely not a comprehensive list, but the list of tools where I can add my perspective. I would appreciate if others can share their experiences as well as in data visualization tools not mentioned here. Thanks!