Statistical Forecast Options

Forecast Type

There are many different approaches to creating a Statistical Forecast but the most common methods / model types are:

  • Moving Average - Smooths data to create a less volatile line. Smoothed past data can be used to create smoother forecasts.
  • Exponential Smoothing - Similar to Moving Average but uses the forecast error to drive forecast direction.
  • Regression - Uses at least one variable (such as number of stores) to predict another (the volume of sales).
     

Intelligent Systems

Intelligent Systems such as Demantra, JDA Demand, SAP Advanced Planning, Kinaxis, Logility and Infor can automate the analysis of forecast types and their variations and then select the most appropriate. There are two typical approaches:

Best Fit: Apply the most appropriate from a selection that will be used on al combinations (ODP)

Bayesian: Blend the best forecasts across the horizon into one forecast (Demantra)


Forecast Level

The results of a Statistical Forecast will differ depending on the Level that it has been created at. Generally lower is better but issues like short lifecycles and unique customer products can force grouping levels. If you set levels too low the variations in data might be large causing decreased accuracy. Statistical Forecasts generated from higher grouping levels can produce better results than using the lowest levels but inaccuracies can be magnified at aggregate levels! If you set levels to high, the forecast might pick up localised trends and allocate them incorrectly.

Image removed.

The Dimension Levels that you choose should relate to your company data and processes and the needs of the business.

 

Forecast Time

The accuracy of the Statistical Forecast will depend on the length of History analysed by the engine, the horizon of Forecast and the methodology used to created the accuracy measures.  

3 years of past data is recommended to generate decent seasonal trends.  

Engine tuning can dramatically improve forecast accuracy and consider tuning for the horizon that is most critical - for example tune the near forecast for higher accuracy at the expense of distant horizon.

Determine the Calendar (Gregorian, Manufacturing, Fiscal, Composite) but be sure to evaluate the impact of forecasting and reporting across your organisation.  Some calendars cannot have weeks.

Don't neglect the Cycle Time for your forecast.   Do you want to make it faster?  Consider the impact of S&OP / IBP.