Excel 365 help: How to dtermine average of bottom 1% of data?

Soldato
Joined
7 Jan 2007
Posts
10,607
Location
Sussex, UK
I am attempting to do some gaming benchmarks, I have a log of Frames Per Second.

I want to determine what my min 1% value is, I assume this would be an average of the values from 0% to 1%?

How would I go about this?
 
Probably not quite what you're looking for, but i'd just manually organise by FPS low to high, and then calculate how many rows 1% would be, then just do =AVERAGE(A1:A100) or whatever range 1% of total is.
 
You sort of answered your own question:
  • Sort in ascending order.
  • Find the first percentile using a simple method: index = 0.01 * number of items in the list; round to the nearest integer, so 0.33 is 1, for example.
  • Then your measure of average of choice (a1:index). Obviously you can stick your list in any column you like, A's just the first one.
  • Done.
It's not surprising there's an Excel formula for that. Just average it via nesting, and presto. Though don't forget to sort still if you are looking for 'min 1% value'.
 
Thanks guys, I did more fiddling and used conditional formatting to fill in a different colour all the bottom 1% values, it confirms the PERCEBTILE Formula is the one I want.
 
I am attempting to do some gaming benchmarks, I have a log of Frames Per Second.

I want to determine what my min 1% value is, I assume this would be an average of the values from 0% to 1%?

How would I go about this?

Is this for a 1% & 0.1% low reading. If so I'm positive that this is not how it works and is much more complex
 
In places not mathematically sound, too, or so I hear.

lol. mini-necro but yes, it's a bit fishy in places.

You can trip Excel up with 6 numbers, possibly less. Consider the following:

suIDEkb.png


0.3 + 0.3 + 0.3 - 0.3 - 0.3 - 0.3 equals zero. Except if you are using sumifs, in which case it's almost zero but not quite.

Dumb.
 
Average of the bottom 1%? - I'd be surprised if the results were statistically significant pending on how many runs.

No point being that specific unless you have a very large data-set to contend with.
 
Back
Top Bottom