Standard Dev, w/out all data points

[Jdh173]

Disclaimer: I am self taught code monkey, without foundational knowledge. But can make things work... Eventually...

I have three data points that represent average daily usage of a product between that point and the last:

16APR	20APR	23APR
5.09	0.53	0

My table doesn't contain dates in between, only reported volume dates (from which the average daily usage between two points is calculated). In reality the table I need should look like this:

14th	15	16	17	18	19	20	21	22	23
5.09	5.09	0.53	0.53	0.53	0.53	0.53	0	0	0

If I use standard deviation formula, I can only reference the three points I have. But, for a correct, standard deviation I need to reference all points broken out. Is there some clever way for me to achieve this? I created a table in Excel sheets to see what the difference between standard deviation is between the 3 points and the 10 points:

*3 points: Std dev = 2.80
*10 points: STD dev = 2.13

This difference makes a big change in the end calculation I am doing with the standard deviation (inventory reorder point & safety stock calculation).

Any help you can offer would be appreciated! Let me know if more detail needed.

 

[KRice]

It sounds as if the data available are already summarized or reported as "average daily usage"...is that correct? Do you have access to more data? Standard deviation is a metric for characterizing the dispersion of a normal distribution, yet with only three data points (especially the three data points you've described), I don't understand why this distribution should be considered "normal", so finding a standard deviation might be suspect anyway. And even then, relying on a 3-point standard deviation is ill-advised. You mention one idea about generating a larger table, but I don't understand why that would help matters---those numbers are fabricated, right?...they don't necessarily represent the average usage on each of the days represented, so any conclusions based on that expanded set would be dubious.

But I don't really understand the details. For the three points you describe, are the circumstances so different that it can explain why that value is so much larger than the others? For example, are multiple people present and using the product? And the 3rd data point (0), does that represent a day when no one is present, or is everyone still present, but not using the product? Details about the usage (who, when, why, etc.) might be useful in creating a reasonable description of the product's usage distribution, and that could then be used for making projections.