Efficient manner to find a table column

[pm_scorca]

Hi,

inside PQ I've created the A table with some transformations, that loads data from a csv file.

After, I've created the B table as a reference one from A.

Inside the B table I've added a custom column by invoking a custom function that uses Table.SelectRows and Table.Max functions against the A table. It seems that for each custom column value in the B table it occurs a data loading from the A table: the data loading dimension of the B table with the custom column is of some hundreds of megabytes while the data loading dimension of the table B without the custom column is of some hundreds of kilobytes.
I could use a DAX expression to calculate the custom column but I would need to apply some other transformation depending on this column inside PQ and not at a model level.

Any suggests to me in order to solve this issue, please? Thanks

[Vijay_A_Verma]

You can try buffering the TableA before using Table.SelectRows and Table.Max expressions. 

So you can write something like

BuffTableA = TableA

Stepx = Table.SelectRows(BuffTableA.......

[pm_scorca]

Hi,
inside the query for the A table, I've added as a last step something like #"Table buffering" = Table.Buffer(#"previous step", [BufferMode = BufferMode.Delayed]). So, the data loading dimension for the table B is remained the same one as without the custom column, also if it is increased the data loading time.
Using Table.Buffer could be a good solution, but I think to hope using this function in the right manner.
I've said that the A table is created to connect to the csv source and to apply some initial transformations; then I've created the B table as a reference one from A. In this way, I can inside the B table I can to invoke a custom function against the A table avoiding a circular reference. Now, is it right to apply the Table.Buffer function to the A table or is it right to create a new C table as a reference one from A and to apply the Table.Buffer function to the C table (obviously, the custom function has to consider the C table)?

Many thanks

[Vijay_A_Verma]

Without seeing your Query for TableA and TableC, I will not be able to advise on the most optimum solution.