There is an interview on sun.com with Phil Yelland of Sun Labs and Eunice Lee of the Worldwide Operations Business Unit (WWOPS). The two provide details on how Sun Labs helped WWOPS replace its earlier "judgmental" or manual sales forecast with less labor intensive sophisticated Bayesian statistical methods that produce results at least as reliable as the earlier effort.

Sun Labs' RADAR project is at the heart of this change. In the RADAR project researchers use advanced statistical techniques to explore the characteristics of Sun's target markets, with particular emphasis on forecasting for operations management.

RADAR project staff produce sales forecasts for many Sun products on a weekly basis using statistical software they have designed. The results are then passed on to the purchasing and operations staff in WWOPS. RADAR project staff also act as consultants to WWOPS on research and analytical issues relating to statistics and forecasting.

Yelland and Lee stress that the Bayesian statistical techniques used in RADAR are not intended to be a substitute for the forecasting expertise of employees in WWOPS and marketing. Instead, the project's success demonstrates that combining human expertise with statistical algorithms results in forecasts that are better than either approach would produce on its own.

Already, the forecasting techniques at work in RADAR are significantly more advanced than those used on a routine basis anywhere else in the industry. In the future, the project team will continue to work on the development of statistical techniques and models to increase Sun's strategic advantage in this area. [...read more...]

		fullsource
Print Email Current Profile Archive Blog Generate newsletter