Append (Robust) (Model Simulator)
Synopsis
This operator works like the usual Append operator but it keeps all the nominal values even if they are not existing in all of the data sets but only in their meta data.Description
Just like the regular Append operator, this operator adds all the rows of all example sets into one merged set. However, this operator also keeps all nominal values in place even if they are not or no longer existing in all of the data sets. This can be useful to not break models in production since the meta data of the resulting data does not change in this case.
Input
- example set input (Data table)
These ports expects ExampleSets which all need to have the same attributes and value types and will be merged by appending all rows into one large set.
Output
- merged set (Data table)
The merged set of all input data sets where all meta data is still intact even if the values have no longer been a part of the input data sets.
Tutorial Processes
Robust Append for Titanic
This process loads the Titanic twice but filters it down to only first class passengers for the first one and only second class passengers for the second one. You can place breakpoints after the filtering and see that while the example sets have been filtered down, the meta data in the statistics view still shows the existence of the other values with a count of 0.
If we would append those two data sets now with the regular Append operator, the data set would be merged but the information about the existence of the Third class would be lost. However, we are using the operator Append (Robust) to merge them. The result is the same but the meta data stays intact and the Third class is still shown with a count of 0.