Predictors - when to normalize

KNIME Analytics Platform
1

Hi, is there any rule when to normalize data for predictions in KNIME? should it be always for better results?

2

Hey zarniak,

The rules for normalizing data in KNIME are the normal rules as everywhere in data science: Algorithms like k-means (when it just uses Euclidean distance on the input data) are sensitive to normalization (as it says in the Node Description), while others like a Decision Tree are not affected by it.

Best,
Ferry

 

3

Hi Ferry, thanks for answer. is there any systemtic list of where to use and where not need to use normalization? Regards

4

Hey zarniak,

I wasn’t able to find one extensive enough to reference it here, but you can find a couple of short lists on Google. Alternatively you can look up the algorithms as you use them.

Cheers,
Ferry

1 Like