Configure Forecast Models
Overview
Forecast models in GoldFinch are used to calculate forecast quantities for future periods based on historical actual sales data. You can define multiple forecast models, and the Daily Calculations Routines will select the most accurate forecast model with the least errors as the best-fit model for each SKU. The Calculate Forecast routine then uses this best-fit model to project future forecasts.
Best Practices for Forecast Models
Deactivate or Delete Forecast Models with Large Errors: Periodically review forecast models and deactivate or delete those with large forecasting errors.
Clone and Tweak Accurate Models: Forecast models with small errors can be cloned and tweaked, then added back to the active models list.
Try Multiple Models: Since GoldFinch selects the best-fit formula based on accuracy, testing several forecast models is essential to finding the most accurate one for each SKU.
Forecast Model Fields
Field Label | Field Help |
---|---|
Active | Only active forecast models will be used in the forecasting process. |
Calculate First Period Forecast Only | If checked, the forecast quantity for the first future period will be calculated using the best-fit forecast model. The forecast quantity for other future periods will be copied from the first period. If not checked, the forecast quantity for all future periods will be calculated using the best fit forecast model. |
Forecast Model Line Fields
Field Label | Field Help |
---|---|
Adjustment % | Actual sales quantity can be adjusted for the forecast calculation. |
Forecast Model | Links to the specific forecast model. |
Include In Trend | If checked, the period's actual sales quantity will be included when calculating the growth trend percentage. |
Period | The look-back period number (e.g., enter "1" to look back one period before the current period, "2" for two periods back, etc.). |
Seasonal Trend | Specifies the period as Blank, Current Season, or Prior Season. Periods assigned to Current or Prior Season are used to calculate the seasonal growth percentage. |
Weight | The weight assigned to the period when calculating a weighted average. |
Sample Forecast Models
3 Months Moving Average
Period | Adjust % | Weight | Included in Trend | Seasonal Trend |
---|---|---|---|---|
1 | 3 | No | ||
2 | 2 | No | ||
3 | 1 | No |
6 Months Moving Average with Trend
Period | Adjust % | Weight | Included in Trend | Seasonal Trend |
---|---|---|---|---|
1 | 3 | Yes | ||
2 | 2.5 | Yes | ||
3 | 2 | Yes | ||
4 | 1.5 | Yes | ||
5 | 1 | Yes | ||
6 | 0.5 | Yes |
Simple 5% Growth Based on Last Year
Period | Adjust % | Weight | Included in Trend | Seasonal Trend |
---|---|---|---|---|
13 | 5% |
3 Months Moving Average with Seasonal Trend
Period | Adjust % | Weight | Included in Trend | Seasonal Trend |
---|---|---|---|---|
1 | 1 | Yes | Current Season | |
2 | 1 | Yes | Current Season | |
3 | 1 | Yes | Current Season | |
13 | Prior Season | |||
14 | Prior Season | |||
15 | Prior Season |
Forecast Calculation Example
Forecast Model:
3 Months Moving Average with 5% Adjustment, Trend, and Seasonal Analysis.
Period | Adjust % | Weight | Included in Trend | Seasonal Trend |
---|---|---|---|---|
1 | 5% | 3 | yes | Current Season |
2 | 5% | 2 | yes | Current Season |
3 | 5% | 1 | yes | Current Season |
13 | Prior Season | |||
14 | Prior Season | |||
15 | Prior Season |
Scenario:
Today's Date: 3/25/20
Inventory Period Type: By Month
Current Period Date: 3/1/20
Actual sales for prior periods:
Period | Period Starting Date | Actual |
---|---|---|
1 | 2/1/20 | 10 |
2 | 1/1/20 | 20 |
3 | 12/1/20 | 30 |
13 | 2/1/19 | 10 |
14 | 1/1/19 | 11 |
15 | 12/1/18 | 12 |
Forecast Model Calculation Template 032520.xlsx
Steps to Calculate Forecast Quantity
Calculate Adjusted Quantity for each period:
Period 1: 10 + 5% = 10.5
Period 2: 20 + 5% = 21
Period 3: 30 + 5% = 31.5
Calculate Weighted Average:
(31.5+42+31.5)/(3+2+1) = 17.5
Calculate Average Trend:
(-0.5+0.33)/2 = 17%
The maximum growth trend is 200% per period
Calculate Forecast Quantity with Trend %:
17.5 x (1+17%) = 14.58
Calculate Seasonal Trend
Current Season Total = 10+20+30 = 60
Prior Season Total = 10+11+12 = 33
Seasonal Growth % = (60 - 33) / 33 = 82%
Calculate Forecast Qty. with Season Trend
14.58 * (1+82%) = 31.82
The final forecast quantity for this period is 31.82.