General
Below is a brief description of the different software suites and features that can be found within this user guide:
Inventory Suite
AGR is an Inventory Planning and Optimization solution that minimizes both inventory costs and shortages. It uses historical data to determine the forecast, which is calculated at item level for each location. The system then dynamically calculates order proposals from these forecast values and many other inputs. The order proposals can be adjusted in the AGR solution and when accepted transferred into ERP system. The Inventory suite includes a ’light’ version of the Planning suite, which can be found within the Plans tab, and is used for general sales and promotions planning.
Forecasting Engine
This manual describes an overview of the statistical techniques, statistics, and strategies that are implemented in the forecasting engine. It is not necessary that you fully understand, or even read, this manual in order to produce accurate forecasts with the product. Included in this manual is the short version for regular users. The long version for advanced users is available upon request. The information included in this manual provides a way for users with little statistical knowledge to find what they are looking for.
In the ever-changing business environment, it is essential to keep an ongoing review of the forecasting process. Automatic forecasting enables your company to reduce the time spent selecting forecasting methods, as the forecasting engine dynamically selects the most appropriate forecasting model to suit the sale pattern for your products. When the sale pattern changes the forecast changes with it.
The forecasting methods used in the Forecasting module can be grouped into the following five categories:
- Simple moving average
- Exponential smoothing
- Croston’s intermittent demand
- Discrete distributions
- Box-Jenkins (ARIMA)
The automatic selection is determined by which method gives the least forecasting error; keep in mind that every forecasting model available will produce errors, since while the world is changing the future will remain unpredictable.
To estimate the lowest forecasting error, thus the most fitted forecasting model, the forecasting engine does a repeated search through the historic data to minimize the sum of squared errors.