Entry

Math: Statistics: How can you test a sequence of random numbers? Chi square test

Oct 3rd, 2004 13:17
Knud van Eeden,

----------------------------------------------------------------------
--- Knud van Eeden --- 03 October 2004 - 09:12 pm --------------------

Math: Statistics: How can you test a sequence of random numbers? Chi 
square test

---

Steps: Overview:

1. generate a sequence of random values using your routine

2. compare this sequence with a standard sequence

   e.g. -look at the square of the differences between your values and
         their values, and add this.

         Thus create the sum of this values:

             ( ( observed value ) - ( expected value ) )^2
             ---------------------------------------------
                          ( expected value )

         another way of writing this is:

( (# value occurs) - ((total measurements) . (chance value)) )^2
-----------------------------------------------------------------
     ((total measurements) . (chance value))

        -compare this sum of this to values in standard tables.

          -If your sum is smaller then you assume your values are
           random

          -If your sum is equal then you assume your values are random

          -If your sum is bigger then you assume your values are not
           random

3. The total degrees of freedom equals the total of your observed 
values
   minus 1.

   E.g. you made 1000 measurements, then you need to look for 1000-1 =
        999 degrees of freedom.

---
---

Internet: see also:

---

Math: Number: Random: Link: Overview: Can you give an overview of 
links about random numbers?
http://www.faqts.com/knowledge_base/view.phtml/aid/31689/fid/1712

----------------------------------------------------------------------