Page 1 of 1
Residual sum of squares in validations
Posted: Wed Feb 16, 2005 4:08 am
by bhaskarac
Dear All,
I have a doubt with the calculation of residual sum of squares in linearity studies.Can any body explain how to calculate in linearity study or is there any literature available on net?
Thanks in advance,
Bhaskar
Posted: Wed Feb 16, 2005 8:05 pm
by redzilla
Hi,
I entered my data into MS Excel spreadsheets. When you perform a linear regression on your data, part of the output is the residual sum of squares. I don't recall the actual formula to calculate it by hand, but if I find something I will post it. Spreadsheets have saved me a lot of time with some of my calculations. =)
Posted: Wed Feb 23, 2005 9:57 pm
by Sallybeetle
I'm with redzilla. I tried to do the rsq calculation by hand
ONCE 
.
Found the function in Microsoft Excel and haven't looked back since.
Posted: Thu Feb 24, 2005 12:49 pm
by bhaskarac
I got the following calculation. Please correct me if I am wrong.
Replicate 25% 50%
1 155 242
2 149 247
3 155 257
X 459 746
n 3 3
x bar(Average) 153 248.6666667
Sigma x square 70251 185622
Sigma x whole square 210681 556516
sigma x whole square/n 70227 185505.3333
Residual sum of square 24 116.6666667
The difference between sigma x square and sigma x whole square/n is nothing but RSS.
Also please explain me what is the significance of the RSS(residual sum of squares) in linearity study?
Bhaskar