Trend
A trend denotes the systematic fluctuations in demand for an item, both upwards and downwards, from one forecast period to the next. Trends can be used as a factor or quantity.
Description
Trend estimates are used when extrapolating base forecasts.
Calculation method
Described methods are used to calculate a trend factor and trend quantity.
Calculating the trend factor
The trend factor is defined as the mean relation (ratio) between demand in two sequential calculation periods minus 1. This calculation is based on trend periods of user-specified length. This formula is used to calculate the trend factor:
Tf(i + 1) = (Dr(i) + .... + Dr(i–x / 2 + 1)) / (Dr(i–x / 2) + ...... + Dr(i–x))–1
A negative trend factor signifies a downward trend, and a positive trend factor signifies an upward trend.
Calculating trend quantity
Trend quantity is the mean difference in demand during two sequential forecast periods. This formula is used to calculate the trend quantity:
Tq(i + 1) = ( * (F(i + 1)–F(i)) + (1–() * Tq(i)
Key:
| ( | = | Smoothing constant for exponential smoothing |
| Dr(i) | = | Actual demand during period (i) reduced by any manual forecast adjustment and non-representative demand |
| F(i) | = | Base forecast during period (i) |
| Tf(i) | = | Trend factor for period (i) |
| Tq(i) | = | Trend quantity for period (i) |
| i | = | Period number |
| p | = | Trend calculation period length |
Description
Assume this demand for an item:
| Jul. 95 | Aug. 95 | Sept. 95 | Oct. 95 | Nov. 95 | Dec. 95 |
| 120 | 109 | 117 | 118 | 121 | 125 |
Base forecast Dec. 95: 119
Base forecast Jan 96: 124
( factor used: 0.3
Number of trend calculation periods: 12
This trend factor and trend quantity are calculated for January 1996.
Calculating the trend factor
Tf(Jan.) = (Dr(Dec.) + Dr(Nov.) + Dr(Oct.)) / (Dr(Sept.) + Dr(Aug.) + Dr(Jul.))–1 = (125 + 121 + 118) / (117 + 109 + 120)–1 = 364 / 346–1 = 0.05
Calculating trend quantity
Tq(Jan.) = 0.3 * (124–119) + (1–0.3) * 3 = 0.3 * 5 + 0.7 * 3 = 3.6