Businesses make and sell a wide range of products to a diversity of customers globally. Each product, customer and geographic region have unique demand patterns. Generating an accurate forecast for every combination of product, market and region is the responsibility of the demand planner (of the organization). The demand planner cannot use a one size fits all forecasting process for all combinations. Neither can they determine the best forecasting approach for each combination manually.

Arkieva’s statistical forecasting engine employs a proprietary best in class “Best Fit” algorithm to identify the best statistical forecast method for each combination. The best method is selected from hundreds of methods (in Arkieva) include moving average based, trend based, cycle based (for example seasonal or Fourier) and more special methods such as ARIMA (Box-Jenkins), causal and Croston’s method.

The statistical forecasting engine is also integrated with segmentation and life cycle management modules. As a result, the output of those modules can be used to drive the forecasting engine differently for different segments of the data.

Built on Arkieva’s attribute based planning data model, the statistical forecast engine can be run at multiple levels of product (or product related attributes – hierarchy or family), market (or market related attributes – ship to, sold to, market, channel), location (or location related attributes – region, country) and kept reconciled via dynamic top-down and bottom-up reallocation.