Weibull Net in KNIME

Hello everyone,

i hope somebody already tried to do what i would like to have: A Weibull-Net in KNIME.

I work with the statistical distribution calles Weibull distribution. It's often used within survival analysis to give back a probability for surviving: F(x) / or the risk to die: 1-F(x) depending on a variable x wich can simply be years a person is old or in the case of a technical product the hours it has been used or something similar.

There is a transformation for this distribution wich is often used cause is transforms the Weibullcurve to a straight line which makes it easier to fit a distribution to a statistical estimator.

To get that line you have to transform the lifetime by: ln (x) and the probability to survive by: ln ( -ln (F(x) ) OR ln ( ln ( 1/F(x) )

There is a simple example attached which shows the curve on the left and the transformed line on the right. The problem is that you can't interpret the right graph that easily cause the values aren't that easy to read as on the left. There you have the direct value of x and a corresponding probability in percentage for example. I would like to have an interactive graph in KNIME which give me back the line on the right but shows the normal values on the axis like on the left. That would look something like the second attachment and is often called a weibullnet.

So far i use a simple scatter plot graph with a HiLite Filter, so i can select a few points and use those selected values for a linear Regression to estimate the parameters of a well fitting weibull distribution. That looks kind of like the attached example of a weibullnet with some points and a fitted line. But for the selection of the points it would be really helpful to see immediately which range of x i selected.

I hope i was able to explain it well enough. Google didn't help so far.

Maybe somebody has an idea to get to that or has already done it.

Thanks a lot.


This sounds like the sort of thing that you will want to do in R.  Have you tried our R integration out yet?

Have a look at:




I think those two pages should provide the information you need.