Polynomial regression: Excel vs. Rapid Miner

Jens_Fleckenste

Hi All,

I tried to understand the polynomial Regression within Rapid Miner.

I know this from Excel and there it works like this:

For example I have a Polynom
y = 1*x^3+2*x^2+3*x+20

xValues:1,2,3,4,5,6,7,8,9,10
yValues:26,42,74,128,210,326,482,684,938,1250

This is the Training Set.

The test Set with the same x_Values
xValues_New:1,2,3,4,5,6,7,8,9,10

should return the same prediction for Y.

In Excel the Function
=TREND(yValues;xValues^{1.2.3};xValues_New^{1.2.3})
returns the expected values 26, 42, 74, 128,.........

The function =RGP(yValues;xValues^{1.2.3})
returns the coefficents of the Polynom
1,2,3,20

Is it possible to rebuild this in RapidMiner with getting the same results?

I setup this process with completly different results. Can someone explain me, what's wrong?

By the way, the same stuff with a linear function or quadratic function works. (Is a function third degree to much for Rapid Miner?)

<参数键= " attributes_lower_bound”value="-10.0"/>
<参数键= " attributes_upper_bound" value="10.0"/>

Thomas_Ott

I think the operator you're looking for the is the Local Polynominal Regression operator.

Jens_Fleckenste

Thanks a lot.

Thats exactly what I needed!

:-)

Kind regards

Jens