Question : Finding a Percentile with array of numbers

I want to find out what number is the nth Percentile of a group of numbers.

For example:

If I have the numbers (1,2,3,4).  What number would be in the 75th percentile?

The answer is 3.25.   I can easily do this calculation by using the "Percentile" function in excel.  Anyone have any idea to do this with SQL?

Answer : Finding a Percentile with array of numbers

The percentiles are nasty thing to do in SQL...

1) What is percentile 75:
Percentile divide ordered data into portions. Percentile 75 is a value from set of data, that 75% of the data values are equal to or less than.

For example percentile 33 from the data (1,2,3,4) is 2.

The algorythm is described here: http://www.staffs.ac.uk/sands/buss/bscal/mandev/m_qm/t_mld/mld_o.htm , but in brief:

For percentile P
1) order the data nad count them (n - number of values)
2) find the position i = (P*n)/100
3) if i is not a whole number, the percentile is the next value in the ordered data (on i+1 position)
4) if i is a whole number, the percentile is calculated from values on i and i+1 position - the calculation, as I know, should be an average from these two values, but it looks like the Microsoft thinks some other way ;-)

Anyway the right way to do this in SQL is following:

A) get the count of the data and store it in a local variable
B) calculate the 'i' value from the formula above
C) test if the i is a whole number
D) do a select over the requiered column ORDERED BY this colum ASCENDING
E) declare cursor over this result sets
F) in a loop do fetch to get to the i-th (and if the i is a whole number to (i+1)th value)
G) if the i is a whole number calculate the percentile from the i and i+1 value
H) return the result

As I said - it is a messy thing to do in SQL - It will do a table scan over whole table ;-(

P.S. you can also modify this algorythm to scan the table from the bottom values (for percentiles <= 50) and from the top (percentiles >50)