Hi, I'm beginner of knime. I want check similarity between two or more record of dataset. The problem is that a subset of attribute are categoricals and another subset are numerics.
I thought:
The dataset is compared with cosine similarity for categoricals attributes, on the other hand, numerics attributes of each row is compared with euclidean distance. And so I have two distance matrix one in cosine and one in euclidean.
Now, how I can normalize and sum two distance matrix?
did you already see our distance matrix plugin? There is a node, called the Java Distance, which allows you to define you own distance. Second there is also a Aggregated Distance with which you can combine two distances into one.
Here is the blog post https://www.knime.org/blog/distance-measures-in-knime which we published together with those extensions.