I want to take a small data set like the one below and calculate the difference between each feature as either a matrix or as a 1:1 readout
Input
Matrix output
Paired output
the goal would be then to use this data to evaluate a matrix of points to determine their proximity in relationship of the nodes for the given point
Hi @jnotter The FeatureJoiner is great for comparing features together. I'd also recommend renaming the attributes for the Left or Right port. See https://hub.safe.com/publishers/safe-lab/templates/pair-every-feature for reference.
Either option is possible.
The matrix option is a bit more complicated as would either require amending to expose the node names as attributes or doing this dynamically. What format would you want to write back to.
I've attached a quick example
difference_between_features.fmwt
Either option is possible.
The matrix option is a bit more complicated as would either require amending to expose the node names as attributes or doing this dynamically. What format would you want to write back to.
I've attached a quick example
difference_between_features.fmwt
I couldn't tell from the transformers how you made the Cartesian join successful, but your FMWT worked as designed once I made my input match your input structure.
Hi @jnotter The FeatureJoiner is great for comparing features together. I'd also recommend renaming the attributes for the Left or Right port. See https://hub.safe.com/publishers/safe-lab/templates/pair-every-feature for reference.
Feature Joiner requires a "Join On" criteria. Essentially this needs a Cartesian join, and thus why the dataset must remain small in order for this to work.