I need the linear regression coefficients for A->B, A->C… A->Z, B->A, B->C… … Z->A to be computed. ie: each attribute must become the label and then have the linear regression coefficients be computed for each of the other attributes.
These are to be computed as simple linear regression and not multiple linear regression.
Given the amount of attributes, I just can't do this by hand.
This is not as simple as in the case of weights because we have no port that delivers these values. You have to dive into RapidMiners APIs with the "Execute Script" operator like in the case of the parameters of a decision tree: http://rapid-i.com/rapidforum/index.php/topic,7834.0.html
Thanks fras. I am not familiar with Java but I can make it out. Where could I see how I can access the different parts (coefficient of an attribute and intercept) of the output of the linear regression? I did not find any documentation that was that technical. i.e.: where could I see what the equivalent of j48.measureTreeSize() and j48.measureNumLeaves() from your example (below) would be called?
Answers
what exactly do you mean by "between all of them"?
Best regards,
Marius
Suppose I have attributes, A, B, C…Z
I need the linear regression coefficients for A->B, A->C… A->Z, B->A, B->C… … Z->A to be computed. ie: each attribute must become the label and then have the linear regression coefficients be computed for each of the other attributes.
These are to be computed as simple linear regression and not multiple linear regression.
Given the amount of attributes, I just can't do this by hand.
Thanks for your reply!
Please have a look at the attached process, it does the job for you
Best regards,
Marius
That is just perfect because I can pull that in excel and have a ton of regressions figured out in an instant, which is exactly what I need.
I have m, how could I extract B in the same table then? (b from: Y = mX + b)
You have to dive into RapidMiners APIs with the "Execute Script" operator like in the case of
the parameters of a decision tree:
http://rapid-i.com/rapidforum/index.php/topic,7834.0.html