How to Caculate Accuracy

[SandMan86]

I am trying to determine how accurate a forecast is.

For example, a customer requested 18 items. However, they only pick-up or take 6 items. The forecast of their needs is only 33.3% accurate.

• 18 forecasted
• 6 actual
• 33.3% accuracy

Calculating a formula that can handle the reverse scenario is more difficult.

• 16 forecasted
• 25 actual
• 44% accuracy (right?)

I used the following formula to calculate the "Accuracy" column below. However, you can see that it returns -%'s. Does anyone have a better idea about how this might be done?

=1-((STDEV.P(0, (A2-B2)*2))/A2)


[TABLE="class: grid, width: 500, align: center"]
<TBODY>[TR]
[TD]Forecast
[/TD]
[TD]Actual
[/TD]
[TD]Accuracy
[/TD]
[/TR]
[TR]
[TD]16
[/TD]
[TD]16
[/TD]
[TD]100%
[/TD]
[/TR]
[TR]
[TD]14
[/TD]
[TD][/TD]
[TD]0%
[/TD]
[/TR]
[TR]
[TD]14
[/TD]
[TD]29
[/TD]
[TD]-7%
[/TD]
[/TR]
[TR]
[TD]17
[/TD]
[TD]5
[/TD]
[TD]29%
[/TD]
[/TR]
[TR]
[TD]19
[/TD]
[TD]17
[/TD]
[TD]89%
[/TD]
[/TR]
[TR]
[TD]11
[/TD]
[TD]26
[/TD]
[TD]-36%
[/TD]
[/TR]
[TR]
[TD]20
[/TD]
[TD]17
[/TD]
[TD]85%
[/TD]
[/TR]
[TR]
[TD]20
[/TD]
[TD]16
[/TD]
[TD]80%
[/TD]
[/TR]
[TR]
[TD]29
[/TD]
[TD][/TD]
[TD]0%
[/TD]
[/TR]
[TR]
[TD][/TD]
[TD][/TD]
[TD][/TD]
[/TR]
</TBODY>[/TABLE]

 

[Jonmo1]

Welcome to the board.

Perhaps

=1-ABS(A2-B2)/A2

 

[SandMan86]

Jonmo1,

Thanks for the quick response! I tried out your suggestion, and it provides the same result -- the result above of -7 and -36 still shows to be negative. I think it has to do with the "1-" part of the formula, which I was using, as well. These particular rows are more than twice the forecasted amount, perhaps that's why. Maybe I can build an IF function to correct it.

Thanks, again!

 

[Jonmo1]

I think the negatives come when the actual is more than double the forcast..

I'm not sure how to handle that..

Why do you not want the 100%+ percentage when the actual is higher than the forcast?
This would be just a standard B2/A2

 

[Jonmo1]

Can you post your 'expected' results given the sample table you provided?
And describe the math steps you used to achieve those desired results.

 

[SandMan86]

The desired result is to achieve 100% forecasting accuracy vs actual-- which means that the variance between the forecast and actual is equal to 0. From that perspective, my goal is to monitor how far from 0 the forecast and/or actual numbers are, and that is why I started with a standard deviation formula.

Originally, I was monitoring the inaccuracy in terms of whole numbers. I.e., the forecast was 10, the actual was 8, the variation was 2. But, this doesn't make a whole lot of sense from a performance standpoint, over time. So, I thought, "I'll just take the percentage, and that'll be fine." But, when the actual is higher than the forecast, the percentage doesn't tell me the accuracy of the forecast so much as the percentage of the actual.

I've been trying to figure out the math end of it, and I think I might be doing the math completely backwards for the forecast / actual. Here's what I've been working on:

(5 (Forecast) - 20 (Actual)) / 5 (Forecast) = -3

(5 (Forecast) - 20 (Actual)) = -15 (or, just 15 from 0) so 5 / 15 = 33.3% accuracy


I've totally confused myself, now... Let me know if this makes any sense! I appreciate your time!