M Query Table.Group with Min based on two columns - excel

Question

I have a table with many columns. Three of these columns are:
Package Name (text)
Units Required (Int.64)
Assessment (Int.64)
What I am trying to do is to find the 'Minimum' "Package Name" first by selecting the smallest number of "Units Required", then because sometimes there are several instances where the number of required units will be the same, the row with the lowest "Assessment".
I am exploring the Table.Group() approach but I am not getting anywhere with my understanding of it. I am doing this in Power Query in Excel 365.
Psuedo Code would be something like:
Table.Group("Previous Step Name",{"Package Name"},{MIN("Units Required"),MIN("Assessment")})
As an aside - is it possible to use a single Table.Group and group at two levels? such as "Package Name" and "Column X" so that the result would be a: for each "Package Name" then for each "Column X" in each "Package Name" (nested as it were).
Thankyou in advance for taking a look at this.
Any help greatly appreciated.
Cheers
The Frog

Answer 1

I think you have to do it step by step.
Data
Queries
Load_Data
Load data from Excel table
let
Source = Excel.CurrentWorkbook(){[Name="Table1"]}[Content]
in
Source
Min_Unit
Identify min unit by grouping with empty "group by" field.
let
Source = Load_Data,
Group = Table.Group(Source, {}, {{"Min_Unit", each List.Min([Units Required]), type number}})
in
Group
Min_Unit_And_Assessment
Use inner join to filter original data for entries which equal min_unit. Next, group by "units required" to get the min_assessment.
let
Source = Table.NestedJoin(Load_Data, {"Units Required"}, Min_Unit, {"Min_Unit"}, "Min_Unit", JoinKind.Inner),
Group = Table.Group(Source , {"Units Required"}, {{"Min_Assessment", each List.Min([Assessment]), type nullable number}})
in
Group
Result
Inner join to filter original data for the combination of min_unit and min_assessment.
let
Source = Table.NestedJoin(Load_Data, {"Units Required", "Assessment"}, Min_Unit_And_Assessment, {"Units Required", "Min_Assessment"}, "Min_Unit_And_Assessment", JoinKind.Inner),
RemoveUnnecessaryColumns = Table.RemoveColumns(Source,{"Min_Unit_And_Assessment"})
in
RemoveUnnecessaryColumns
Result

Answer 2

Qualia, thankyou for pointing me in the right direction.
The way that I solved this was really simple in the end!
Step 1: Sort the rows based on the grouping criteria (package name, system class) in that order
Step 2: Add an Index Column so each row has a unique ID to work with
Step 3: Group the table based on the same fields (package name, system class) and 'aggregate' on the lowest Index Number (MIN)
Step 4: Perform a 'Merge Queries' with a Left Outer Join using the Index Number as the matching field between your current 'step' and the step from earlier in the processing where the Index was added - you can then have the rows matched and only the rows needed will be matched since the others are now gone due to the MIN aggregation from earlier. Here is my example:
Table.NestedJoin(#"Grouped Rows", {"Winner"}, #"Added Index", {"Index"}, "Lookup Data", JoinKind.LeftOuter)
- Grouped Rows was the grouping step (Step 3)
- Winner is the name of the Index that had the minimum value
- Added Index was the last step before grouping that still had all the columns (Step 2)
- Index is the column that was added after the sort to uniquely number each row
Step 5: Expand the table and select the columns of data that you want to hang onto
Treating it a bit like a database was a good approach and I appreciate the suggestion you put together for me. Hopefully this will allow others to solve some of their problems too.
Cheers and many thanks
The Frog