Oracle Inventory Forecasting

April 25, 2014/in News /by Trinamix Team

Oracle Inventory Forecasting

Inventory forecasting is the process of extrapolating the expected demand of an item over a number of periods in the future.

Forecasts generated by Oracle Inventory are based on historical transaction activity only. When creating a forecast, you can select the type of transactions that you want to use. You can also specify how far into the future that you want to forecast demand.

After you complete a forecast, you can use it to determine reorder-point planning. You can also load forecasts into the master schedule, which is a component of the Oracle Material Requirements Planning (MRP) application.

Note: If you install only Oracle Inventory, you can manually create forecasts. Manually created forecasts can be based on transaction activity other than historical data

Describing Forecast Types:

Forecast generation uses mathematical algorithms to calculate a prediction of future demand. You can calculate estimated future demand for items using historical data and focus or statistical forecasting techniques. You can create multiple forecasts and group complimenting forecasts into forecast sets. Oracle Inventory supports the following forecast types:

Focus

Statistical

Focus forecasting enables you to simulate various methods of calculating demand so that you can select the best forecasting model. Statistical forecasting enables you to use detailed history and applies weighting factors to exponentially smooth the data. Statistical forecasting also enables you to apply exponentially weighted trend and seasonality factors to predict demand.

You typically use Focus forecasting to produce single period forecasts, whereas you can use Statistical forecasting to forecast any number of periods into the future.

Describing Focus Forecasting Methods:

Focus forecasting simulates five forecast methods to determine the best forecasting model to use. An example of each focus forecast method is shown in this slide and on the following pages. Each method generates a forecast for the current period based on demand from previous periods.

Note: If not using daily time buckets, focus forecast methods 1 and 4 require at least one year of historical data. When you use daily time buckets, a week is used instead of a year in calculating Models 1 and 4. Fifty-two week years are presumed in yearly calculations with weekly time buckets. This means that the same week last year is taken to be the week fifty-two weeks before the current week.

1)Forecast = actual demand in the same period of the previous year

Example: Demand for April 2000 = Demand April 1999

2)Forecast = actual demand in the previous period this year

Example: Demand for April 2000 = Demand for March 2000

3)Forecast = (actual demand in previous period this year + actual demand two periods ago this year) / 2 Example: Demand in April 2000 = (March 2000 + February 2000) / 2

4)Forecast = actual demand in the same period last year × (actual demand in the previous period this year / actual demand in the previous period before the same period last year)

Example: Demand for April 2000 = Demand in April 1999 × (March 2000 / March 1999)

5)Forecast = actual demand in the previous period this year × (actual demand in previous period this year / actual demand two periods ago this year)

Example: Demand for April 2000 = Demand for March 2000 × (Demand for March 2000 / Demand for February 2000)

Determining the Best Forecast Method:

The system uses the absolute percentage error (APE) to determine the best forecasting method to use. The APE is the difference between the actual demand and the forecast associated with the actual demand. You calculate the APE using actual and forecast demand. Oracle Inventory selects the model with the smallest APE to calculate the current period forecast.

The following formula determines the APE:

APE = ( |actual demand – forecast demand| ) / actual demand

Note: Focus forecasting provides a one-period forecast. If you request a focus forecast for multiple periods, then Oracle Inventory uses the forecast of the first period for all of the forecast periods in the request. If actual demand is available for the current period, then you can recompile the focus forecast to update the forecast.

Describing Statistical forecasting:

Statistical forecasting uses exponential smoothing toextrapolate demand from previous periods. Thestatistical forecast methods that you can use with Oracle Inventory include the following:

Exponential smoothing (ESF)

Trend-enhanced forecast (TEF)

Season-enhanced forecast (SEF)

Trend- and Season-enhanced forecast (TSEF)

Describing the Exponential Smoothing Forecast (ESF):

Exponential smoothing uses the forecast from the prior period and adds an adjustment to obtain the forecast for the next period. With ESF, demand is forecast by averaging all of the past periods of actual demand. This forecasting method weighs more recent data to give it greater influence over the forecast results than older data.

You can calculate the current period forecast by using a weighted average of the most recent and forecasted demand. The alpha factor, also called the smoothing constant, is multiplied by the forecast error to determine the adjustment. You can specify an alpha factor between zero and one. The larger the alpha factor, the less impact the older data has on the new forecast.

The current forecast is equal to the old forecast, plus a portion of the forecast error from the previous period. You can use this method when trend or seasonality patterns do not exist.

Example of Exponential Smoothing Forecasting:

This example shows the ESF calculations for three different values of alpha a for period 9 of a 9 period time frame.

The forecast for period 9 for a = 0.9 is calculated:

ESFt = a × At-1 + (1 – a) × ESFt-1

ESF9 = 0.9 × A8 + (1 – 0.9) × ESF8

= 0.9 × 270 + 0.1 × 288

= 271.8

As shown in the table, actual demand for period 3 was abnormal, but otherwise the the trend is upward. With a higher alpha the forecast reacted more strongly to the third period, and produced a very low period 4 forecast, but was also faster to correct itself and adjust for the trend. By period 9, the period that this example is forecasting, the abnormal period 3 has only a minor effect on the forecast. All three forecasts become more accurate when they have more historical data upon which to draw.

Note: ESF always lags behind the trend by at least one period.

Describing the Trend-Enhanced Forecast (TEF):For longer-range forecasts, you can use the trend-enhanced forecast to estimate the amount of persistent change in basic demand from period to period.

TEF is based on the exponential smoothing factor (a), but also considers the trend (b).

Both the exponential smoothing and trend values closer to zero are weighted towards the past trend and values closer to one are weighted more heavily towards the current trend.

Example of Trend-Enhanced Forecasting:This example shows the effect of adding the trend enhancement to the ESF calculation. With the Trend-enhanced forecast, you can reflect the current trend in a forecast.

Assuming an alpha (a) value of 0.5 and a beta (b) value of 0.1, The trend-enhanced forecast for period 9 is derived by performing the following calculations:

Determining base value

Updating the trend index

Adding the two for the period 9 trend-enhanced forecast

Determining the Base Value:

Bt =a × At-1 + (1 – a) × TEFt-1

B9 = 0.5 × A8 + (1 – 0.5) × TEF8

= 0.5 × 270 + 0.5 × 306

= 288

Updating the Trend Index

Rt = b × (Bt – Bt-1) + (1 – b) × Rt-1

R9 = 0.1 × (B9 – B8) + (1 – 0.1) × R8

= 0.1 × (288 – 287) + 0.9 × 19

= 17.2

Adding the Base and Trend Index to Determine the Period 9 Trend Forecast

Recall that the calculated base value was 288 and the trend index was 17.2.

TEFt = Bt + Rt

TEF9 = B9 + R9

= 288 + 17.2

= 305.2 = Period 9 Forecast

Describing the Trend- and Season-Enhanced Forecast (TSEF):The Trend- and Season-Enhanced forecast, combines the trend and seasonal methods to incorporate both types of demand. With TSEF, you specify a trend factor, as well as a seasonality index.

Example of Trend- and Season-Enhanced Forecasting:

Despite the seasonal adjustments made in the SEF, a trend element remained as seen in the gradual increase in the forecast base, B. The TEF and the SEF can be combined to derive a trend- and season-enhanced foremast (TSEF). The TSEF uses all three smoothing factors: alpha (a), beta (b ), and gamma (g ).

As with the Season-enhanced forecast, you calculate the new period 8 seasonality index as soon as period 8 actual demand is determined.

To calculate the trend- and season-enhanced forecast for period 9, you perform the following calculations:

Calculate the period 8 seasonality index

Calculate the period 9 base value

Calculate the new trend factor

Add the base and trend factors together and multiply by the seasonality factor to get the period 9 trend- and season-enhanced forecast

For this example, assume the following values:

Smoothing constant, a = 0.5

Trend smoothing constant, b = 0.1

Seasonality smoothing constant, g = 0.3

Calculating the Period 8 Seasonality Index

S’t = g × [At / (Bt + Rt)] + (1 – g ) – St

S’8 = 0.3 × [A8 / (B8 + R8)] + (1 – 0.3 ) – S8

= 0.3 × [270 / (255 + 10)] + 0.7 – 1.15

= 1.11066

Calculating the Period 9 Base Value

Bt = a × (At / S’t-1) + (1 – a) × (Bt-1 + Rt-1)

B9 = 0.5 × (A8 / S’8) + (1 – 0.5) × (B8 + R8)

= 0.5 × (270 / 1.11066) + 0.5 × (255 + 10)

= 254.04935

Calculating the New Trend Factor

Rt = b × (Bt – Bt-1) + (1 – b) × Rt-1

R9 = 0.1 × (B9 – B8) + (1 – 0.1) × R8

= 0.1 × (254.04935 – 255) + 0.9 × 10

= 8.90494

Calculating the Period 9 Trend- and Season-Enhanced Forecast

TSEFt = (Bt + Rt) × St

TSEF9 = (B9 + R9) × S9

= (254.04935 × 8.90494) × 1.10

= 289.24972

Describing Forecast Sets:

Before you define forecast rules and forecasts, you should first define a forecast set. Forecast sets group together complimenting forecasts. The forecast set also holds a number of parameters that are applicable to all forecasts in the set. Note: A forecast can be associated with only one forecast set, although multiple forecasts may be associated with one set.

How to Access the Forecast Sets Window:

Inventory Responsibility (N) Inventory > Planning > Forecasts > Sets

How to Set Up a Forecast and Forecast Sets

(Help) Oracle Inventory > Inventory Planning and Replenishment > Forecasting

The Defining a Forecast window opens. To set up a forecast set, access the Prerequisites section, and click the “Defining a Forecast Set” link for detailed setup instructions.

Defining Forecast Rules:

Before you generate a forecast, you must specify the forecast rules. Forecast rules, define the content of your forecast. These rules include specifying the following information:

Rule name and description

Bucket type (buckets specify the time period in which your forecast refers. This time period refers to the times periods that you set up when you set up the organization calendar).

The demand sources, such as sales order shipments

Forecast definition information, such as the forecast type (focus or statistical)

Alpha and trend factors

Seasonality factors, if required

How to Navigate to the Forecast Rules Window:

Inventory Responsibility (N) Inventory > Setup > Rules > Forecast

How to Set Up Forecast Rules:

Use the following navigation path and instructions to access instructions on how to set up forecast rules.

(Help) Oracle Inventory > Inventory Planning and Replenishment > Defining a Forecast Rule

The Defining a Forecast Rule window opens.