Reusable SAS/GRAPH Stored Processes for Dashboarding

SAS/GRAPH gives the SAS developer unparalleled flexibility for creating compelling custom data visualizations. At the same time, data visualization is becoming more important in creating Web dashboards. By combining SAS/GRAPH and the stored process HTTP level interface, the SAS developer can create visualization stored processes that are data neutral, reusable, and dashboard ready. This approach yields stored processes that can be applied immediately to any data and be used from almost any type of Web application. Most importantly, the visualization stored processes encapsulate complex SAS/GRAPH code and provide an interface so that developers can use them without any knowledge of SAS programming. INTRODUCTION In 1815, American general and future president Andrew Jackson fought and won the Battle of New Orleans. But the entire battle was fought after the War of 1812 had ended. Jackson and his British counterpart suffered from a lack of timely information because the news of the end of the war was still crossing the Atlantic. But in today’s world of cheap, nearly limitless disk space and instantaneous always-connected communication, decision makers and experts rarely suffer from a lack of information. Instead, the enterprise is awash in information—much more information than can be absorbed and processed by the decision makers and experts in your organization. Moving forward 186 years from the end of the War of 1812, the Enron case presents the opposite information problem. Enron carried on the perception of an innovative, profitable company with spectacular revenue growth for years before reporters uncovered the truth—Enron was actually losing money at a staggering rate and its stock valuation was suspiciously high. A few months after reporters initially reported that the Enron stock was overvalued, the company was bankrupt, partially because the company’s charade was dependent on an ever-rising stock price. But the reporters weren’t lacking for information to reach their initial conclusion—they reached it by analyzing the publicly available filings published by Enron in accordance with securities laws. Enron wasn’t a case of an outright lack of information, but analyzing the right information in a sea of data. Web dashboards, a relatively new type of reporting, are meant to address the problem of information distillation. They provide executives and others with at-a-glance overviews of key metrics within the contexts of their expected behaviors. With SAS, you can both distill down to the key metrics as well as visualize the data. Dashboarding is most effective in your organization if the development of dashboards is both cost-effective and customized for your organization’s need. This paper outlines an approach to SAS/GRAPH stored process development that results in reusable visualization modules. The visualization modules are also data neutral, interchangeable with other visualization modules that adhere to the same contract, and embeddable into almost any Web application. This paper initially looks at the different dimensions of dashboarding and illustrates how SAS/GRAPH provides dashboards that are customized, data rich, and that have a high impact. Next, the basic architecture of dashboarding stored processes is reviewed. Moving from there, the paper focuses on developing stored processes in accordance with two principles: separation of presentation logic and data logic, and the use of a contract approach to achieve flexibility and reusability. The penultimate discussion demonstrates how to make the stored process visualization modules embeddable in any application. This paper focuses on stored processes as a real-time mechanism of delivery for visualizations. In many cases, realtime isn’t necessary and might be a performance detriment to the overall system. Fortunately, the practices suggested in this paper can apply to standard SAS programs that simply publish images on a routine basis. While this paper doesn’t discuss the batch approach in detail, this paper concludes with some pointers about how to apply these approaches to non-stored process SAS programs. DASHBOARD DESIGN AND SAS/GRAPH “Dashboard” is one of a long line of metaphors in computing—others include desktop, file, folder, and even socket. The dashboard metaphor has been successful because it relates to something most of us see every day in our cars. In the car, the dashboard provides information about key metrics in context. You know what the fuel level is and how close it is to being critically low. The dashboard metaphor is so compelling that many early dashboards took it too far and tried to make enterprise dashboards look exactly like a car dashboard, including the windshield, rear view mirror, and steering wheel. Applications Development SAS Global Forum 2007