[Feedback] Compare of limited use

Hi

I just compared my exercise workflow to the solution workflow calling the context menu item, because I cannot make my results get near the proposed solution. But to me, this compare is of very, very limited use if the comparison takes into account the node ids instead of the workflow structure. Do I miss something?

Kind regards

Thiemo

My solution

Proposed solution

Comparison
image

The comparison feature is used to see changes in your workflow over time. Comparing two completely different workflows is quite hard.

I would guess your settings are different in the learner node

1 Like

I thought so about the purpose, but if you delete a node and recreate it later because you can/did not want to revert to a previous version, because you then needed to re-implement the big rest, you are done for. What about comparing two versions of different environments, e.g. one workflow behaves unexpectedly and one wants to find out whether the version of another environment was deployed correctly?

It was my guess too, that I some settings were different. After comparing screenshots of the settings dialogues and still not finding the source, I remembered the compare command. I happened to compare the output of the predictor node and found no differences in the samples I looked at. So, now I believe, the problem is in the scorer though I did not see any config difference.

If you want to, upload your workflow so we can check it

1 Like

I just thought of comparing exports or files behind the nodes.

If I put my teacher hat on, it is good you cannot just find the solution but must try it. This is better for understanding it :wink: :wink:

Doo, I was just feeling stupid not to be able to upload to my hub in the public section. I did it be context menu, I tried by drag n drop, nothing seemed to happen, nothing appeared. Some time later, I was told that it was too big…

Sorry, I need to leave for half an hour.

is the reset version.

@Iris If I put my pupil’s hat on, I, many times, fiddle around with settings to see what happens. :smiling_imp: :grin:

You are using the wrong column in the Numeric Scorer You are comparing to the variance and not to the predicted value:

image

1 Like