Turn column names in M Language defined by variables

[M001]

Hi there, 

I would like to check the data in my latest month column against previous month, and return Y for rows with data in the last two columns not equal

The tricky part is the number of columns are adding every month so the column name for the latest month and the month prior are changing every month, e.g Latest will be 1/6/2024 when in June and it will change to 1/7/2024 when in July

My approach was:

1. Identify the latest month and the prior month >> turn them into variable as LatestMonth and PriorMonth. I have to make the latest month and prior month as variable because the latest month is changing every month

LatestMonth = 1/6/2024
PriorMonth = 1/5/2024

2. Replace [#"1/6/2024"] to LatestMonth and [#"1/5/2024"] to PriorMonth in the following code

= Table.AddColumn(FilteredStep, "RecentUpdate", each if [#"1/6/2024"] = [#"1/5/2024"] then "N" else "Y")

= Table.AddColumn(FilteredStep, "RecentUpdate", each if LatestMonth = PriorMonth then "N" else "Y")

Unfortunately, it doesnt work, and hope someone could enlighten me. Thank you!!

Regards,

[dufoq3]

Hi @M001, check this:

Result

let
    Source = Table.FromRows(Json.Document(Binary.Decompress(Binary.FromText("i45WclTSUTIEYiMgNgZiEyB2dFSK1YlWcoJKwaRB2NUVLOWMJAyjnZzAUi5IhhlD2c7OYClXJDtg2MVFKTYWAA==", BinaryEncoding.Base64), Compression.Deflate)), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [Column1 = _t, #"1/3/2024" = _t, #"1/4/2024" = _t, #"1/5/2024" = _t, #"1/6/2024" = _t, Column2 = _t]),
    Ad_RecentUpdate = Table.AddColumn(Source, "RecentUpdate", each 
        [ a = List.LastN(List.Sort(List.Select(Table.ColumnNames(Source), (x)=> (try Date.From(x) otherwise null) is date)), 2),
          b = if Record.Field(_, a{0}) <> Record.Field(_, a{1}) then "Y" else "N"
        ][b], type text)
in
    Ad_RecentUpdate

[m_dekorte]

Hi @M001,

Copy this M code into a new blank query, replacing its content with:

(t as table) as table =>
let
    cols = Table.ColumnNames( t ),
    dates = List.Transform( cols, each 
        try Date.FromText(_, [Culture="en-GB"]) otherwise null 
    ),
    names = Table.MaxN( 
        Table.FromColumns(
            { cols, dates }, 
            type table[ Fieldname=text, Value= any]
        ), "Value", 2 
    )[Fieldname],
    addCol = Table.AddColumn( t, "Recent update", each
        if Record.Field(_, names{0}) = Record.Field(_, names{1}) 
        then "N" 
        else "Y", type text
    )
in
    addCol

Rename this new query to: fxRecentUpdate

Now you can invoke it on your table by selecting it from the drop down.

Note. This does require at least two "date field" column names to be present else it will return an error.

I hope this is helpful

[foodd]

Hi @m_dekorte Melissa, and thank you for helping to solve this practical problem with PowerQuery/M.

This problem would be a great addition to your LinkedIn. Today Menu series.

[slorin · Accepted solution]

Hi @M001 

if List.IsDistinct(List.LastN(Record.ToList(_),2)) then "N" else "Y"

Stéphane 

[M001]

Thank you, Stephane and everyone. I think this is the simplest approach among all the helpful feedback, just that the N and Y were placed wrongly, and we need to make sure the two columns must be at the last two position

[WanderingBI]

If you hardcode the column names your query will be very hard to maintain in the long run.

Normally it is better to first get your input data into a more "standard" data table layout (columns for field names and rows for values). The reason is that all the functions in Power Query are more aimed at a layout like that. In your case it might also be better to not just to a lot of adjustment steps in Power Query (like transposing the table), but to instead adjust your data source (as an example an Excel file) in the first place. Column names are really not the right place to store transactional data.

Month	Value
2023/01/01	3
2023/02/01	10

If you really are after months it would also be advisable to to transform your dates to months, or months and years.