Remove Other Columns - problems

[PetyrBaelish]

As part of a reporting solution, I get an extract of data from an IT system that contains about 50 columns - but I only need about 8 of them. This is a CSV file that is saved on SharePoint Online. When connecting to the data I (first of all connect to a SharePoint folder and filter down to just thew required csv file & 'expand the data to see the columns within the csv file, then) transform the data to keep only the 8 required columns by selecting the 8 columns followed by "remove other columns".

I have since discovered a problem whereby the column names seem to change upon downloading the data from the IT system, so where the original columns were named...

keepA, keepB, keepC, keepD, keep1, keep2, keep3, keep4, remove1, remove2 ... remove42

they are now:

keepA, keepB, keepC, keepD, keep1.0, keep2.0, keep3.0, keep4.0, remove1.0, remove2.0 ... remove42.0

and on the 3rd attempt, there was a different amount of decimal places and the behaviour is very sporadic.

So I have, in summary:

A) 4 columns that I want to keep whose names remain the same

B) 4 columns that I want to keep whose names change

C) 42 columns that I want to remove whose names change.

Point A above is fine, point B above I accept that every time I update the CSV I will have to manually re-transform the data so that Power BI wil detect the new column names. But for Point C, as I have selected "Remove Other Columns" I don't want to have to rename 42 of them just to remove them straight away - I would have expected the names not to matter as my action is to keep selected columns and remove all other columns (i.e. Power BI is focussed on what is being kept, not what is being removed)

This is a solution for end users to maintain - I don't want to have to tell them to remove these 42 columns on the csv file or do any transformation work as this will be a cumbersome task for them - renaming the 4 columns to keep (point B) is as much as I can get away with.

Is there any way I can get around this in Power BI. As the report developer I want to implement something that means all the end users need to do when refreshing the data is manually rename 4 columns in the CSV file and that's it.

[v-eachen-msft · Accepted solution]

Hi @PetyrBaelish ,

You could remove this step in the query editor.

[PetyrBaelish]

Thank you. This has worked. By having the changed type step before the "remove other columns" step - there was a reference to all of the columns specific names which is what caused the problem.

[d_gosbell]

Is the position of these columns always the same? If they are you could remove the promote headers step from the query, then remove the first row and manually name the columns you want to keep in order to make sure that the column names do not change. You could also use the trick in this blog post https://datachant.com/2017/01/24/power-bi-pitfall-5/ so that the names of the columns you do not want to keep are not important.