--- In

caostheory@yahoogroups.com, "gandhi_gaurav_nsit"

<gandhi_gaurav@...> wrote:

>

> I have recently conducted a specific test on a sample of 1200 items

> and output of each item was between 0-100. Is there any way (or

> theory) by which I can know where is my system biased toward or if my

> system of items is chaotic or random ?

>

> What should be the minimum sample size to test the randomnes

>

Take one of your sets of data (100 samples) and find the minimum member:

use that as the unit scale and make a set of integers

out of the sample sizes. Then you count the frequency of each sample

size.

You plot the log of the time the sample was taken against

the frequency of that sample.

You call the slope of that plot your next level of data.

You do that to maybe 1/4 of you 1200 items

and plot item number against slope of the log-log frequency plot.

If the line that results is near a straight line at slope two,

the results would be called random.

If the average is some other number, then the results can be called

"fractal".

This is the Zipf type word frequency test.