Thank you for the 2.0! It’s really the best! But I have a problem, when I try predict values.
I have a table, where some of the values are missing. Totaly, table contains 50 000 rows, with five nominal columns (one of them is target). I divide this table into traning (about 35 000 rows) and target parts by using “Row Filter” node and try to use WEKA nodes. I filter both traning and target samples by “Reference Row Filter”, so it contains same values there, I think. But, when I execute “WEKA Predictor”, with Ridor, BF Tree, and some others I have error “Value not defined for given nominal attribute!”. Whats mean this error?
This error message occurs when there is a difference between the domain of the training data and the domain of the test data. Usually, this should not happen.
Is it possible for you to send me your flow (with data)? nicolas.cebron(at)uni-konstanz.de
How can I estimate the domain of applicability with the KNIME tools?
I found this on a the web:
but I don't know to implement this with the R-snippet node. I also wonder how KNIME deals with the visualisation.
KNIME includes a simple Statistics node for univariate statistics, but also allows using all R (statistics) functions from within the R Snippet and R View node (for visualizations). Those nodes a available through our update site. Just try it out.
I have the R extension installed.
Now I wonder how I get it technically running... I have a training set with some descriptors, does this correspond to the y1 values?
And the test set corresponds to y2?
But there is only one input flow port - how should this in practice?
Thanks in advance!