When trying to make a simple decision tree it happens that I get decimal values for count and bad coloration. it occurs with features having missing values and disappears when I replace missing values.
The problem is that I have no alert so :
- it can lead to bad decision
- it has no sense…
- it has a bad impact the ability for my collegues to trust Knime’s result
Here a dataset (use H_fistule) as response variable in the tree.
decimals are visible when you expand the tree
fistule.csv (91.4 KB)
using KNIME 5.1.2