Spring 2006 Issue

Special Feature: Forecasting for Call Centers

  • Introduction A call center is a centralized office where inbound calls are received – normally from customers requesting assistance – and outbound (telemarketing, survey) calls are made. Often staffing hundreds of agents, the call center concentrates the telephone-based hardware, service, and support in one location, a configuration that entices many companies to outsource their telephone functions.
  • Nano Forecasting: Forecasting Techniques for Short-time Intervals by Jay Minnucci
    Call centers and other organizations that deal in real-time environments must be able to forecast in days, hours, and even minutes. They can do so successfully by finding smaller bits of data hidden within the “macro” data. Jay shows how this nano-forecasting focus can be employed to project call volumes and to improve resource productivity.
  • Forecasting Call Flow in a Direct Marketing Environment by Peter Varisco
    Peter provides a case study in the use of dynamic modeling to forecast call volumes and to estimate how these volumes are affected by the timing of direct mail campaigns. Dynamic modeling, variously called dynamic regression, ARIMAX, and transfer function modeling, is a driver-based (explanatory) methodology that can supply precise timing effects of key drivers such as direct mail promotions. In summarizing the lessons from the application of this methodology at New York Life Insurance, Peter provides a working demonstration of the method’s value for call-volume forecasting.
  • Forecasting Weekly Effects of Recurring Irregular Occurrences by Dan Rickwadter
    Weekly forecasts are important for call centers, but they present a host of challenges, including “recurring irregular occurrences” such as paydays and billing cycles. In this article, Dan describes his techniques for cleaning the weekly data, accounting for the irregular-event effects, and generating weekly forecasts of call volumes.

    • Commentary by Tim Montgomery

Special Feature: Forecast-Accuracy Metrics for Inventory Control and Intermittent Demands

  • Introduction by Editor, Len Tashman
    This special feature of Foresight examines the challenges in measuring forecast accuracy within the context of inventory management. We give special attention to the difficulties in accuracy measurement that arise when demands are intermittent.
  • Measuring Forecast Accuracy: Omissions in Today’s Forecasting Engines and Demand-Planning Software by Jim Hoover
    Forecast-accuracy metrics are critical guidelines for proper selection and implementation of forecast models. In demand planning, improved accuracy and better modeling translate into reduced inventory costs, increased service levels, or some combination thereof.
  • Forecast-Accuracy Metrics for Intermittent Demands: Look at the Entire Distribution of Demand by Tom Willemain
    While most forecast-error metrics are averages of forecast errors, Tom argues that, for intermittent-demand series, we should focus on the demand distribution and assess forecast error at each distinct level of demand. He illustrates how this can be done, and he suggests use of the chi-square statistic to judge the overall effectiveness of the forecast method.
  • Accuracy and Accuracy-Implication Metrics for Intermittent Demand by John Boylan and Aris Syntetos
    John and Aris distinguish between forecast-accuracy metrics, which measure the errors resulting from a forecast method, and accuracy-implication metrics, which measure the achievement of the organization’s stock-holding and service-level goals. Both measurements are important. The correct choice of a forecast-accuracy metric depends on the organization’s inventory rules and on whether accuracy is to be gauged for a single item or across a range of items. The authors recommend specific accuracy and accuracy-implication metrics for each context.
  • Another Look at Forecast-Accuracy Metrics for Intermittent Demand by Rob Hyndman
    Some traditional measurements of forecast accuracy are unsuitable for intermittent-demand data because they can give infinite or undefined values. Rob Hyndman summarizes these forecast accuracy metrics and explains their potential failings. He also introduces a new metric—the mean absolute scaled error (MASE)—which is more appropriate for intermittent-demand data. More generally, he believes that the MASE should become the standard metric for comparing forecast accuracy across multiple time series.


    1. Managing the Introduction of a Structured Forecast Process: Transformation Lessons from Coca-Cola Enterprises Inc. by Simon Clarke
      Simon Clarke, Manager of Forecasting and Planning for Coca-Cola Enterprises Inc., led a corporate team that engineered a radical transformation of the forecasting process. The team took his organization from an unstructured, decentralized process to a disciplined internal collaboration of over 2,000 forecasters in a highly volatile promotional environment. Here he describes the lessons learned in managing the transformation.
    2. Breaking Down Barriers to Forecast Process Improvement by Mark Moon
      Mark draws upon his experience in audits of the forecast process at many large companies to identify the key barriers to forecast process improvement and how these barriers may be overcome. He examines the critical role of the forecast audit; discusses the need for changes in organization structure, forecast process, computer systems, and performance measurement; and explains how a forecast champion may be necessary to successfully implement the requisite changes.
    3. Tips for Forecasting Semi-New Products by Bill Tonetti
      “Semi-new” is the label that author Bill Tonetti assigns to products that are not truly new, but rather result from extensions and modifications of existing products. In this article, Bill shows how to use data that exist on the predecessor products to forecast demand for the semi-new products. Many firms overlook the opportunities Bill describes here, with severe consequences for forecast accuracy, inventory costs, and service levels.
    4. Lessons from Thomas Edison’s Technological and Social Forecasts by Steven Schnaars
      Thomas Edison’s inventions have had an unparalleled influence on modern life. But Edison was also a technological forecaster, offering his vision of which technologies would (and would not) dominate our lives in the future. Steve Schnaars looks back on Edison’s 13 technological and social forecasts to evaluate the inventor’s predictive hits and blunders. The main lesson he sees in Edison’s technological forecasting is that spreading risk by pursuing multiple paths to future market success is probably a better strategy than trying to predict precisely which technologies will succeed in the future..
    5. Book Review by Anirvan Banerji
      Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets by Nassim Nicholas Taleb


Go to Top