Expression for Standard Deviation with different variations.

[Jeremy Wiedenmann]

I’m looking to write an expression looking for the Standard Deviation of a group of rates. The result needs to exclude values that are zero and that are negative from this calculation. The ultimate goal is to get the correct standard deviation results for each group and the visualize the top 10-15 groups that have a standard deviation +/- 2 times the standard deviation. I’ve been able to get pieces of this, but not everything as one. Any help would be appreciated. Thank you.

[David Hall 5]

It has been a while, but there are some commands available at the FOCUS code level that used to generate Standard deviation.

How are you attempting to calculate the Standard Deviation?

[Jeremy Wiedenmann]

Right now just using the expression StdDev([frate]). Frate is a water amount that is bringing back values of 797.15 to -15.5 and such. So if the StdDev for the frate of this meter is 1,352 over the last 3 months, it should actually be 986 for those 3 months because there were negative meter volumes as well as volumes of zero that I need to exclude.

[David Hall 5]

So how are you filtering out the negative or negative values.

It would be best to filter the data into a hold file and then do the standard deviation off the hold data.

[Jeremy Wiedenmann]

They are not filtered yet. I'm still wanting to show them in the data, but excluded from the calculation, if possible.

[David Hall 5]

Filter out the data you want into a hold file first.

Then work off that hold file on the calculation.

Once you get the calculation correct, then you can go back to doing the filtering on the fly.

Statistics are very specific and someone not knowing the data can screw things up

For example if you have a grade point average, you cannot average the average.

You cannot take a B average from one semester and a C average from a 2nd semester and get an average.

One MUST take all the grades, and the points for the grades and then recalculate the average for all the semesters.

Standard deviation needs the individual values to get the correct number

Another thing you may want to do is work on a small sample of data first.

If you have TIBCO's Rstat, they have SD as one of the tab options to compare it to.

[Jeremy Wiedenmann]

That has helped a lot. Thank you!