Is there a way to define data as categorical or numerical in rapidminer Go, or is it done only automatically? Also, is there a way to define a column as an ID so it is not taken as data into the model, but is out put unchanged in the results?
Dear jsdrew, We are working on a data preparation and exploration feature that will allow you to make these changes. Would you be open for a feedback call with product and design? Regards, Andras
Hi alebo, By any chance the data preparation and exploration feature is ready to release? I'm also looking for feature described by jsdrew, especially a way to define a id column which will not be used in the model training, but need to be retained in the output, so that it can be used to identify some properties of the record or join with other data.
Hi @RionArisu, We are working on a feature that will detect potential data quality issues and let you fix them. This should cover some data preparation use cases (including setting roles for certain columns). Our plan is to have a beta release in late-May.
Regarding ID columns, we've recently pushed out a new version of Go. It now keeps unused columns when scoring a model. This should let you use the ID to join with other datasets. Here is a screenshot to demonstrate the behavior:
Answers
We are working on a data preparation and exploration feature that will allow you to make these changes. Would you be open for a feedback call with product and design?
Regards,
Andras
By any chance the data preparation and exploration feature is ready to release?
I'm also looking for feature described by jsdrew, especially a way to define a id column which will not be used in the model training, but need to be retained in the output, so that it can be used to identify some properties of the record or join with other data.
We are working on a feature that will detect potential data quality issues and let you fix them. This should cover some data preparation use cases (including setting roles for certain columns). Our plan is to have a beta release in late-May.
Regarding ID columns, we've recently pushed out a new version of Go. It now keeps unused columns when scoring a model. This should let you use the ID to join with other datasets. Here is a screenshot to demonstrate the behavior:
Regards,
Andras