If you choose to use the "First" value in the Aggregation section, you can choose to take the first value even if it is a missing value (i.e. when the Missing box is ticked). Taking the tick off only takes the first value it comes across which isnt missing. This works on all the columns, except when the column in Smiles. The GroupBy node continues to take the missing value even when it has been unticked.
if the missing value option is unticked the first operator only returns a missing value if a group contains only missing values independent of the data type. It has to return a value even if the group contains only missing value.
Yes agreed, but the data set does contain a value!
This is really odd. Does each group by itself also contain a value? To check this, could you please add an additional count without missing values aggregation opertation that uses the same column you perform the first operation on in order to ensure that all groups have at least one value. The result column should have for each group at least count greater zero. If this is the case could you please send me a small workflow that contains the data and the groupby node. If the data I confidential can you give me some more details on the data such as how many columns, data types, rows, groups, etc. and the settings of the groupby node so that I can try to reproduce the problem because I have tried it with several data sets and settings and could not reproduce the problem.
I have eventually, with some effort, pinned down where the problem is.
It appears to be one of our internal nodes, which is filling the missing value cells with a "?" string rather than an actual missing value. Therefore, obviously, the GroupBy node thinks the string "?" is a real value, hence it is reported even when "First" aggregation is selected with no missing values.
Apologies for this, I will need to discuss this with our programmers.
no problem at all. I'm glad it wasn't my fault :-)