Forum Discussion
split, process and recombine data
- 4 months ago
Your dataset is silent on the actual FTE values so I will make an assumption that the FTE values can be related to an employee via Employee ID and are available in an external table.
With that assumption in place, you can get your desired result by selecting the 'Department', 'Prev Row Department', and 'Employee ID' columns and then unpivoting based on 'Department' and 'Prev Row Department'.
From there you can merge in the FTE values from the external table and then turn the 'Prev Row Department' rows negative.
Here is a sample code,
let Source = Table.FromRows( Json.Document( Binary.Decompress( Binary.FromText( "hdJdb4IwFAbgv3LCtQmFgs5LP2aMHwuZizfGi6rH0Qg0Aczifv1aZmudxd215PDkPT1ns/FI4HW8mUgLWOW8TuWFEp90/ZCEXXl5g6k4VyhPU3MKqE+oLgj7pnrbkVyoOFbgHRfp6oVB5rAUonRykcVRxWFV8T37FUGRgfXHEtYcv+RhCCNR1+zzwVQJqGVG146Lyy2kFGNdn2jxI0VIWFlleHGZsWXGTc5vtsPMjQ5M56/uh/wrqm8znrs103QCE5Ed/rN6zVBOLUO5RXtvn7E9lJem2db3Gz+Gi56E6181fsTSHXAIE1bmz3fQyheQpt39iaN7qUeGWemorTu9/QE=", BinaryEncoding.Base64 ), Compression.Deflate ) ), let _t = ((type nullable text) meta [Serialized.Text = true]) in type table [#"Employee ID" = _t, #"Full Name" = _t, #"Effective date" = _t, Department = _t, #"Prev Row Department" = _t, #"Prev Row Effective date" = _t, #"Prev Row End effective date" = _t] ), initial_type_set = Table.TransformColumnTypes( Source, { {"Employee ID", Int64.Type}, {"Full Name", type text}, {"Effective date", type text}, {"Department", type text}, {"Prev Row Department", type text}, {"Prev Row Effective date", type text}, {"Prev Row End effective date", type text} } ), unpivot_selected = Table.Unpivot( Table.SelectColumns( initial_type_set, {"Department", "Prev Row Department", "Employee ID"} ), {"Department", "Prev Row Department"}, "Attribute", "Value" ), merge_fte = Table.NestedJoin( unpivot_selected, {"Employee ID"}, fteTable, {"Employee ID"}, "fteTable", JoinKind.LeftOuter ), expand_fte = Table.ExpandTableColumn( merge_fte, "fteTable", {"FTE"}, {"FTE"} ), set_negatives = Table.ReplaceValue( expand_fte, each [FTE], each if [Attribute] = "Prev Row Department" then -[FTE] else [FTE], Replacer.ReplaceValue, {"FTE"} ), remove_columns = Table.RemoveColumns( set_negatives, {"Employee ID", "Attribute"} ), final_type_set = Table.TransformColumnTypes( remove_columns, { {"FTE", type number} } ), group_rows = Table.Group( final_type_set, {"Value"}, { {"FTE", each List.Sum([FTE]), type nullable number} } ), rename_value = Table.RenameColumns( group_rows, { {"Value", "Department"} } ) in rename_valueI used the following code to create FTE values based on the example data you provided.
let Source = #table( type table [#"Employee ID"=nullable number, FTE=nullable number], { {1,.5}, {2,1}, {3,1}, {4,.8}, {5,1}, {6,1}, {7,.66}, {8,1}, {9,1}, {10,1} } ) in SourceTo end up with the result. (Note: I grouped by 'Department', summing the FTE.)
Hi AndrewPF
This isn’t actually a transpose problem—it’s about splitting each row into two records to represent movement in and out. The most effective way to handle it is in Power Query by creating two rows per original record: one for the “Loaned To” department (using the Department column with a positive FTE value) and one for the “Borrowed From” department (using Prev Row Department with a negative FTE value). You can do this either by duplicating the query and appending, or more efficiently by creating a list column that generates both records per row and then expanding it. This approach ensures you end up with the required 20 rows, correctly capturing inflow and outflow, and makes your final aggregation by department straightforward and accurate.