How do you initialize the first forecast before applying exponential smoothing?

Conditions

• Used when no prior forecast is available.
• Applied specifically to period 1 in this example.
• After initialization, later forecasts are generated recursively starting from period 2.

Reasoning, step by step

1. Identify the actual demand for the first period, $A_1$.
2. Set the forecast for the first period, $F_1$, equal to $A_1$.
3. Use this initialized $F_1$ as the starting seed for the recursive calculation of $F_2$ and subsequent periods.

Example

In the example, the actual demand for Year 1 is 310. The instructor states, 'Whatever is your actual demand, straight away you can post that as a forecasted demand,' and writes 310 in the Forecast Demand column for Year 1, establishing $F_1 = 310$.

Common misconceptions

• Attempting to apply the exponential smoothing formula to calculate the very first forecast without an initial seed value.
• Assuming the first forecast must be an average of all historical data before the series starts.

Watch the explanation

Explore next

• What is the exponential smoothing forecast formula used in this video?
• Why does the Year 2 forecast come out to 310 in this example?