That’s a great observation, treating those zero values as missing and replacing them with the mode is a really ractical approach. It is a clean and thoughtful workflow, nicely done.
1 Like
Hello @ KNIMErs
Here is mi submission for the JKI S04 CH06 . I still have to do some homework to fully understand all the node configs implications ; then I will get more confident working ML within KNIME.
I’ve added a ‘one-hot encoding’ component from my own, delivered for queries in forum. Within this [ topic ]: You can find the explanation why, it is not applied for Random Forest Learner use case in the submitted WF.
Keep KNIMEing 
2 Likes
Thank you! Coming from an engineer who binge-watched Dr House for all his “medical training,” spotting a flat-line blood-pressure felt like the one diagnosis I couldn’t miss. 
Happy the mode-swap prescription passed your peer-review— no Vicodin 
, just a clean workflow. Looking forward to the next puzzling symptom in the data ward! 
1 Like
We just published our solution to last week’s challenge on heart failure prediction!
This challenge involved parameter optimization and xAI, and provided a good opportunity for honing classic machine learning classification skills.
Come back tomorrow for an AI-powered challenge on soccer! Can you combine your skills and create a report on the players’ strengths and weaknesses?
3 Likes