What method is this?

[hamzah]

Lets say I have three values - 917, 914 and 913. Take an average I get 915 (approx).

 

Now see how far each value is from the average...

 

917 - 915 = 2

915 - 914 = 1

915 - 913 = 2

 

(2+1+2)/3 = 2 (approx)

 

so the value is (915 +/- 2)

 

what is this method called?

[Neurocomp2003]

....method?

this isn't a method its error measurement.

[Kedas]

LOL

 

It does ring a bell of some statistical method with 'some result' but I can't remember.

[hamzah]

Neurocomp2003 said in post # :

....method?

this isn't a method its error measurement.

 

yeah thats what I'm doing, error measurements...but the way I calculated it...does that technique have a name or anything?

[blike]

Thats almost like variance, which takes the sum of all the squares of the differences between the means and values and then divides by the number of samples.

 

In your case, the variance would be:

 

[math]http://blike.com/mimetex/mimetex.cgi?4$var=\frac{(1-2)^2+(2-2)^2+(3-2)^2}{3}[/math]

 

grr..my cgibin sucks

[fafalone]

..standard deviation?

[Glider]

Hamza is talking about the mean deviation.

 

The measure of how much any given observation varies from the mean is a deviation. These will have either positive or negative values. The mean of them will always be zero (as the mean is the arithmetical centre of the data). So, to calculate the mean deviation, we ignore the sign (+ or -) which gives us for each deviation the absolute deviation, shown as |d|.

 

So, the two ways of calculating the mean deviation is either:

 

Sum of |x - mean|

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

             N

 

 

or

 

Sum of |d|

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

      N

 

 

The Standard deviation is: 

 

                                       (Sum of (X - mean)^2) 

The square root of:     -----------------------------

                                                       N

 

 

(or N-1 for the sample SD).

 

Dang! I wish I knew how to use the formula doohickey.

[hamzah]

thanks Glider and the rest of you for your help

[wolfson]

You're working out the mean absolute deviation from the mean, but you really ought to use a more accurate value for the mean of the numbers.

[hamzah]

Well I've completed an experiment to determine the specific heat capacity of Aluminium...and the book quotes "910J/kg.k"...so I thought I'd stick to the same format as them.

[Fluent in Lies]

Diactine Standard Regulative Regression

[wolfson]

Fluent i do not think standard error and effective reduction is applicable at this level.

[Fluent in Lies]

Oh how terribly silly of me, what was i thinking *girlish giggle*

[wolfson]

Its ok