By default, Eclipse uses standard and median forecasting when calculating demand to help you determine what products you need to purchase and how often you need to purchase them. Starting in Release 8.7.8, Eclipse provides additional, or advanced, forecasting methods to help you best determine how to categorize your products and thereby maximizing your purchasing power.
Note:These forecasting methods are usedin addition tothe standard forecasting methods. This process is to complement the original process and help you determine the best forecast to use.
Eclipse analyzes the product history to determine the demand by looking
at minimum hits and history, smoothing factors, and automatic trends.
Use the
This means, Eclipse uses your selected group and goes back the number
of weeks indicated in the
In forecasting inventory, smoothing refers to finding the demand by removing the random variations that may occur in the purchasing history. This helps determine demand patterns more accurately, so that you do not purchase for those exceptional sales or exceptional periods in the life of the warehouse. For standard forecasting calculations, this means taking an average of your sales. With Advanced Demand Forecasting, Eclipse applies base level, trend level, and seasonal level. In formalized forecasting manuals, this refers to Alpha, Beta, and Gamma factors. For ease of explanation we refer to them as Base, Trend, and Seasonal.
Base - Known as the Alpha factor. The recent demand period versus the prior demand period. This field must be set between 0.05 and 0.25. If left blank, the system uses 0.15.
Trend - Known as the Beta factor. Must be set between 0.05 and 0.25. If left blank, the system uses 0.15.
Seasonal - Knows as the Gamma factor. Must be set between zero (0) and 0.5. If left blank, the system uses 0.5. For example, if the seasonal score for this week is 100% of normal, but actual sales were 200% of normal, the forecast adjusts the score up for next year. A higher gamma means a higher adjustment. A gamma of .25 would in that instance have the score as 125% for the following year.
Once calculated, Eclipse automatically calculates smoothing factors
for each item. If the calculation is above the maximum value set in the
Note: Eclipse uses weekly forecasts, meaning 52 weeks' worth of data. Due to this type of calculation, Eclipse limits your maximum smoothing factor values to prevent these factors from including outlying values. This keeps your smoothing factors addressing the most useful data and giving you the best values for managing trends.
The system uses the smoothing factor, the most recent period's calculated demand, and the most recent period's forecast to create and exponential smoothing calculation. This calculation takes into account the previous period and goes back in the demand periods for you and the output represents all the previous period demand. In Eclipse, the default is 24 months of product history, but it can use whatever history your company stores in the Eclipse files.
Using the
In addition, the system can use those exceptional sales and help determine if they truly are exceptional or are actually seasonal product sales. For more information see How Eclipse Test for Seasonality below.
In order for Eclipse to know which advanced forecast method to apply, the system must first test your branches and products in combination for seasonality. This means, the system considers specific parameters or factors to determine if a product should be considered seasonal. By marking a product seasonal limits when you purchase the products.
Eclipse uses the most recent years of history as defined in the
The demand information is gathered four-week groupings culminating in thirteen groups per calendar year. Using this method avoids the influence variations in month length or weekday periods throughout the year. For example February, in comparison to other months, is 7-8% shorter. In addition, months of equal length, such as 30 days, can have an unequal number of weekdays depending on how the months fall in a calendar year. Eclipse then passes the calculated demand value through a Autocorrelation Function (ACF) to determine if the correlation over a one-year period and the 50% confidence level. The system then uses a moving average method for this time series and calculations. If the determined correlation is greater than 0.3 and outside the 50% confidence interval, the item is treated as seasonal.
Note: The seasonal tests do not prove if an item is seasonal, but to show that is more likely seasonal than non-seasonal to have seasonal forecasting methods applied.