SAS Institute Inc. is pushing the limits of forecasting up and out with a new application that extends into the millions the number of products for which it can simultaneously predict demand and that automates more of the analysis tasks so that less-sophisticated analysts can use the product.
The Cary, N.C., company this week will roll out SAS High Performance Forecasting, a module for the SAS 9 Intelligence Platform that performs automated forecasting. The new software includes sophisticated mathematics processing so that users dont need a strong statistical background, officials said.
When presented with a set of historical data, HPF determines which of about 500 statistical models are appropriate for forecasting future demand, runs the calculations, and flags outlying or rogue results for an analyst to monitor.
Todd Kirk has been testing HPF at his company, Ipsos. The company uses the SAS software to forecast the results of “what-if” scenarios for marketing groups within consumer packaged goods and pharmaceutical clients.
“In our world of causal analysis, people are doing what-if games that are all retrospective,” said Kirk in Norwalk, Conn. “What people really want to do, which this tool will enable you to do, is predict if you do X, [then] this is what will happen, and if you do Y, this is what will happen.”
For Kirk, senior vice president of product development, a key benefit of HPF is its ability to provide forecasts within designated hierarchies. “Historically, forecasting is done at the brand level; now, with this, it is done at the SKU level,” he said.
Kirk also welcomed HPFs ability to automatically choose the best statistical model, which allows his analysts to focus on adding more value to the results of the forecast. The softwares ability to crunch millions of numbers quickly was a strong point.
“Maybe thats overkill today, but like anything in technology, as soon as somebody says you can do something, then you are going to find a need that that fills,” said Kirk.