Full breakdown of decision tree

[chrishy]

Dear all,

I have a dataset that contains 8 distinct variables. However, the resulting decision tree brunch out to only 3 categories. How do i achieve a full breakdown of the tree with all 8 variables included?

[ipazin]

Hello @chrishy,

welcome to KNIME Community!

Can you explain a bit more what is troubling you? If you are talking about Decision Tree View from Decision Tree Learner node it can be expanded with plus sing.

If that is not troubling you maybe variables (columns) that are missing are skipped due to no domain information which is usually the case when nominal column has more than 60 unique values.

Br,
Ivan

[chrishy]

Hello Ivan,

Thanks for the reply!
My dataset contains both numerical and nominal columns. However, I noticed that my decision tree (as shown in the Decision Tree View) only shows the numerical columns. My nominal columns does not have more than 60 unique values. Why is this so?

Br,
Christine

[ipazin]

Hello @chrishy,

not sure. You can check workflow Building a Simple Classifier under Basic Examples folder in your KNIME Explorer and see that nominal columns are shown. Can you maybe share your workflow example? (Please don’t share any confidential data.)

Br,
Ivan

[chrishy]

Hi @ipazin,

Below is an example of my workflow:
Predictive Hiring.knwf (15.3 KB)

Br,
Christine

[ipazin]

Hello @chrishy,

can’t check it as workflow is reset. Either don’t reset or include data within your workflow. See here how to do that and more: Reproducible (Minimal) Workflow Example

Br,
Ivan

[chrishy]

Hello @ipazin,

Attached is my workflow with data included. Thanks for the help!
Predictive Hiring.knwf (18.8 KB)

Br,
Christine

[ipazin]

Hello @chrishy,

I see. This is due to how decision tree (algorithm) works. You have small sample so tree simply stops growing If you set minimum number of records per node to one (1) you’ll probably see another (nominal) level in your tree…

Here are couple of thoughts on predictive modelling with small data sets from KNIME Community exprerts:

Small Dataset Predictive Modeling

Hi Community, I am working with a very small dataset (9 rows and 7 attributes). Is it even feasible to train a predictive model with such a small dataset (i.e. decision tree, neural network,etc)? Even if i could train a model, how much weight can I put into the accuracy? Thanks in advance for any feedback!!

And how knows, maybe they join discussion and provide their opinion/ideas on data set provided

Br,
Ivan

[chrishy]

Dear @ipazin,

Do you have any idea what is the minimum sample size needed to attain a complete decision tree?

Br,
Christine

[ipazin]

Hello @chrishy,

don’t think that is the right question to ask. “Is your sample size good enough to get meaningful results from decision tree?” seems a better one. If not there are either other algorithms either methods to boost your sample size.

Br,
Ivan