+6I have a dataset containing 100000 points, for these points I've calulated 700 neighborhood. Now I want to calculate the z score for every point in each individual neighborhood.
My question is: is there an easier way to do it, than 700 conditional values ?
I've tried with a custom transformer in which I can loop, but I cannot get my head around how to do it.
Please help :)
Thanks in advance.
Best answer by jonathanklarup
https://knowledge.safe.com/questions/78585/how-to-add-group-by-to-any-transformer.html
Found it here, so if I put the z-score into a custom transformer I can make a group by parameter which solves the issue :)
+62How exactly are you going to calculate the z-score? If it can be expressed in a single formula I don't think looping or conditional values are even necessary.
700 conditional values are most likely not the way to go by the way, that'll be hell to set up and maintain.
+6How exactly are you going to calculate the z-score? If it can be expressed in a single formula I don't think looping or conditional values are even necessary.
700 conditional values are most likely not the way to go by the way, that'll be hell to set up and maintain.
I calculate the standard deviation and mean with the statistics calculator, and group it by the neighborhood id, then i have the formula in the attributecreator for the z score.
Which i'd like to have in like a for loop: for x in neighborhood : calculate z..
+6How exactly are you going to calculate the z-score? If it can be expressed in a single formula I don't think looping or conditional values are even necessary.
700 conditional values are most likely not the way to go by the way, that'll be hell to set up and maintain.
+6https://knowledge.safe.com/questions/78585/how-to-add-group-by-to-any-transformer.html
Found it here, so if I put the z-score into a custom transformer I can make a group by parameter which solves the issue :)
+6How exactly are you going to calculate the z-score? If it can be expressed in a single formula I don't think looping or conditional values are even necessary.
700 conditional values are most likely not the way to go by the way, that'll be hell to set up and maintain.
https://knowledge.safe.com/questions/78585/how-to-add-group-by-to-any-transformer.html
Found it here, so if I put the z-score into a custom transformer I can make a group by parameter which solves the issue :)
+6https://knowledge.safe.com/questions/78585/how-to-add-group-by-to-any-transformer.html
Found it here, so if I put the z-score into a custom transformer I can make a group by parameter which solves the issue :)
Must have forgotten this part of the advanced course ...
+62https://knowledge.safe.com/questions/78585/how-to-add-group-by-to-any-transformer.html
Found it here, so if I put the z-score into a custom transformer I can make a group by parameter which solves the issue :)
I honestly don't think you even need to group by, the AttributeCreator works feature-by-feature.
+6I honestly don't think you even need to group by, the AttributeCreator works feature-by-feature.
You are completely right. I might have over thinked it .. but GREAT Thanks a lot :)
