• Votes for this article no votes for this yet
  • Dashboard Insight Newsletter Sign Up

One-on-One with Sean FitzGerald
Vice President of Engineering/CTO, Visual Numerics, Inc.

by Karly Gaffney, Media Relations, Dashboard InsightThursday, May 14, 2009

Dashboard Insight recently spoke with Sean FitzGerald, Vice President of Engineering/CTO of Visual Numerics about his company's lengthy history, his definition of embedded analytics and how customers have been using his analytical tools in recent years.



Dashboard Insight:
Tell us about the history of Visual Numerics.

Sean FitzGerald: We have an exciting update to our history just this week. On May 5, Rogue Wave Software announced that it has acquired Visual Numerics. It’s a great fit for our products and will benefit many customers who are developing advanced analytic applications in scalable, distributed computing environments.

Before the Rogue Wave acquisition, Visual Numerics had a long history as a privately held software company.  Visual Numerics actually started as two separate companies: IMSL Company, established in 1970 and Precision Visuals, established in 1980. The two merged in 1992 to form Visual Numerics.  The IMSL Company developed embeddable mathematics and statistics libraries while Precision Visuals developed visual data analysis tools. Merging the two companies and technologies made good sense as many customers who are analyzing data need to visualize the results and vice versa.   

We’re proud of our long history and we’re looking forward to continuing to work with our customers as part of Rogue Wave.

DI: Where does Visual Numerics sit on the B.I. stack?

SF: Our mathematics and statistics libraries are embeddable, platform agnostic and scalable, so they fit into several places in the BI stack.  At the lowest, or most basic, level - because the libraries work on literally hundreds of platform combinations - they can work in virtually any environment.  Moving up at the data-preparation layer, many customers use our libraries for pre-statistical processing.  And as you go higher in the stack, the libraries can be embedded into databases, BI applications, used in reporting or embedded into any application to add analytic capabilities.  Both the libraries and our visualization product can also extend the analytics of an application as an add-on component.

DI: Visual Numerics calls itself a provider of embedded analytics. How do you define embedded analytics?

SF: At a high level, I think of analytics as an umbrella word that describes using numerical analysis techniques on data to solve problems and make decisions. “Embedded analytics” is simply making the analysis part of an application or process.

The mathematics and statistics algorithms available in our libraries are building blocks used by developers as integrated parts of new or existing analytic applications. Regardless of functionality, whether it’s data mining, forecasting, visualization or something else, the algorithm is embedded in the application; it’s not a wrapper or interface or bridge to the data. We think embedding analytics in this way has its advantages.

Also, the analytics are closer to the data and for most applications, that’s a good thing. Execution and efficiency are improved when you can analyze data at the source instead of moving it into a third-party application for separate analysis.  And multiple users can access analytic capabilities and use the results to make better decisions for their organizations, not just the users who run reports or use desktop BI tools.

DI: Do your customers use embedded analytics in different ways than they did 3, 5 or 10 years ago?

SF: While the basic problem-solving nature of analytics is the same as it was years ago, we’ve seen many changes in how customers use our tools in recent years.

We’ve certainly seen an evolution in the depth and complexity of algorithms that customers are using. Forecasting applications that 10 years ago relied on techniques like simple moving averages now leverage advanced techniques like neural network models. Customers are using the more advanced techniques to either better understand their data or to solve completely different sets of problems.  

Another area that’s significantly changed in recent years is not as much “how” customers are using embedded analytics but “where” they are running analytic applications. The hardware that applications run on today is vastly improved over the hardware of 10 or even five years ago, with multi- and many-core architectures as common today in small systems as they are in higher performance computers. We’ve invested considerable time and effort to enable our algorithms to be part of high-performance applications running on these computers. While high-performance computing might not be a consideration for many dashboard applications today, I expect it’ll become increasingly important as vendors and users add more advanced analytics into dashboard applications.  

DI: In the last three years, are you seeing new types of customers (or customers in industries new to you) being interested in analytics?

SF: In our early years, our customers came from scientific, engineering and research fields. The majority of our users are still in these fields, using our products to do things like simulate fluid flows, analyze flight test data, forecast weather patterns and do other scientific and engineering research.

Recently, we’re seeing more customers coming from the field of business both from organizations wanting to build their own analytic applications and from ISVs wanting to embed analytics so they can provide more analysis capabilities in their products.

The area where we’ve seen the most growth - in organizations building their own applications - is finance. Even in this troubled economy, we’re seeing growth. Optimization and forecasting functions play big roles in analytic applications for our financial services customers. We’re also getting increased interest from ISVs who are including analytic capabilities in their products. SAP, for example, licensed our software in late 2007 to enhance search capabilities in TRex, part of their NetWeaver platform. SAP is obviously a large ISV example but we have many small and medium-sized ISV customers as well. There are examples on our website at http://www.vni.com/company/partners/oemIsvPartners.php.  

DI: Readers of Dashboard Insight are obviously interested in dashboards. Can you provide 1-2 examples of interesting dashboard applications your customers have created?

SF: I mentioned finance as a growing area of activity. A dashboard view that many finance customers use is something called a heat map. A heat map is a chart type that displays two-dimensional arrays of color values. A simple finance heat map might, for example, show data for a set of stocks. Colored rectangles in the map could represent stocks where the price has risen (green), gone down (red) or not changed (white). The amount of green, red or white in the map allows users to see trends that are developing. This type of dashboard is a very simple representation of data in visual form. It's something that you could do with the charting components in several of our libraries or with our visualization product.

Our consulting services group recently completed a more complex dashboard application for a customer that wanted to model and display logistical maintenance costs. The customer manufactures and maintains aerospace equipment so they deal with very large and varied datasets when they pull in real and simulated maintenance data. Their application required dashboard reports to display cost forecasts but also a 3D animation of the simulation data. The 3D animation shows the movement of equipment between locations and the simulated maintenance requirements at each location. For the customer, the 3D view was critical for explaining the simulation process to management and prospective customers.

DI: What advice can you give our readers who want to enhance their dashboards with analytics?

SF: I’ll share three ideas that your readers might consider as a step-by-step approach to thinking about analytics.

First, figure out what question you want answered before spending any time figuring out how you’ll answer it. I know that sounds obvious but we’ve certainly seen examples of customers who’ve added analytic functions because “someone might use them” but then, no one did.

Second, create a prototype model. There are many ways to analyze data. Try using different algorithms on a sample dataset and look at the results. If it’s not clear from the results which method or approach is optimal, ask for help. There are many groups and sources of information on analytics available online and certainly companies like ours offer consulting services to help customers better understand their data and how to analyze it.

Finally, after you have a working prototype model that answers the question you set out to answer in step one, then move that model into your production dashboard.  If you follow these steps and keep users involved throughout the process, you’ll end up with usable analytic components in your dashboards that will help people make better decisions. I expect it’ll also lead to requests for even more analytic capabilities.

DI: What is the process if someone wants to evaluate your solutions?

SF: You can download or request an evaluation copy of any of our products via our website: www.vni.com. Evaluation software is exactly the same as the purchased version; it’s just time-limited.

We offer different products, so what you should evaluate depends on what you want to accomplish and how.  If you’re developing analytic application or analytic components for a dashboard, our IMSL Numerical Libraries provide hundreds of mathematics and statistics algorithms in native C/C++, Java, .NET and Fortran.  Look for the library version in your development language.

If you’re looking for a complete development environment that offers tools for prototyping analytic models and then moving those models into production applications, evaluate PyIMSL Studio. PyIMSL Studio combines easy-to-use Python prototyping tools with IMSL Library algorithms for a complete prototype to production development solution for analytic applications and dashboard components.

And if you’re creating a dashboard from scratch or need to visualize large amounts of data in different ways, PV-WAVE is a 4GL that allows users to obtain data from many sources and visualize it in different ways.

Our product documentation is freely available on our website. If you’re seeking information on a numeric function or wondering what functions you could have available by using one of our products, the doc is a great source of information.

DI: What can we expect to see from Visual Numerics in the coming months?

SF: We’re a software company, so our customers will continue to see new and updated product releases from us.  We’re currently in development of the next release of our Java and .NET libraries. While that work progresses, look for Visual Numerics at trade shows like Microsoft TechEd and SIFMA where we’ll gather more customer feedback to drive the development of future product releases.

You’ll also see us continue to update PyIMSL Studio, adding more algorithms into this product and doing other enhancements to make the product even easier to use. Opening up analytics to a new set of users, such as creators of dashboard applications, is an important part of what we plan to accomplish in 2009.



Sean FitzGerald, Vice President of Engineering/CTO

As Vice President of Engineering at Visual Numerics (now part of Rogue Wave Software), Sean is responsible for leading the development and implementation of strategy and vision for the IMSL Libraries, PV-WAVE Family and PyIMSL Studio.

Tweet article    Stumble article    Digg article    Buzz article    Delicious bookmark      Dashboard Insight RSS Feed
 
Other articles by this author

Discussion:

No comments have been posted yet.

Site Map | Contribute | Privacy Policy | Contact Us | Dashboard Insight © 2017