APO DP - Forecast Model Parameters: First-Order Exponential Smoothing

[marianoc]

This document reviews the impact of the Parameters for the Constant Forecast Model (First-Order Exponential Smoothing). The explanation follows a graphical approach without going into the heavy mathematics behind them.

1. Model Selection: Sales pattern

If you want to select a model manually, then you must first of all analyze past consumption data to determine whether a distinct pattern or trend exists according to which you can manually select a model for the system.

Constant requirements pattern:

If your past data represents a constant consumption flow, you can then select either the constant model or the constant model with adoption of the smoothing factors. In both cases, the forecast is carried out by first-order exponential smoothing.

You have another two possibilities if your past consumption pattern is constant; either the moving average model or the weighted moving average model.

2. Forecast Model Parameters: First-Order Exponential Smoothing Models

APO calls this method "Constant", because the resultant forecast is constant.

Model Parameters:

Alpha factor:

The system uses the alpha factor for smoothing the basic value. If you do not specify an alpha factor, the system will automatically use the alpha factor 0.3.

In the following forecasting graphs you will see the impact of Alpha:

a = 0.1

a = 0.2

a = 0.3

a = 0.4

a = 0.5

a = 0.6

a = 0.7

a = 0.8

a = 0.9

a = 1.0

The choice of Alpha is very important: the following table contains the monthly forecasting quantity of the different a alpha factors:

a	Monthly Forecast
0.1	172191
0.2	140316
0.3	116096
0.4	97226
0.5	83583
0.6	73978
0.7	67247
0.8	62622
0.9	59673
1.0	58213

Note that the forecast got with alpha 0.1 is 4 times bigger than the forecast got with alpha 1.0.

Historical Periods:

In the table you will see the impact of changing the historical periods with alpha = 0.3:

a	History	Monthly Forecast
0.3	6	120331
0.3	12	118608
0.3	18	116065
0.3	24	116096

Conclusion: 0.3 is a good compromise

• a (alpha) close to 0: use more historical data smooth the historical data.
• a (alpha) close to 1: use only the most recent data the historical data are not smoothed the ex-post forecast is lagged