A tool desk recorded 18, 21, 23 and 26 checkouts over four weeks. Estimate week 5 with a straight-line fit, then check what the method would have predicted before week 4 arrived.

Enter the observations and fit one future point

Put week numbers 1 through 4 in A2:A5 and the checkout counts in B2:B5. In a separate cell, enter:

=FORECAST.LINEAR(5,B2:B5,A2:A5)

The estimate is 28.5. Keep the fractional result visible when checking the model. If a plan needs whole units, make the rounding choice separately.

Four observed weekly counts are 18, 21, 23 and 26; a fitted line estimates 28.5 for week 5.

Illustrated forecast. The point at week 5 is an estimate in the same unit as the observed counts.

Withhold a recent value and check the error

Use =FORECAST.LINEAR(4,B2:B4,A2:A4). This uses only weeks 1 through 3 and gives about 25.67 for week 4. Compare it with the observed 26. A close result in one check gives context for this example. A changing schedule or seasonal pattern can make the next estimate less reliable.

The held-out check compares an estimate of 25.67 for week 4 with its actual count of 26.

Illustrated sanity check. The held-out estimate uses only weeks 1 through 3.

Microsoft’s FORECAST.LINEAR function reference.