Forum Discussion
A generalisation of the MAP lambda helper function
- Sep 08, 2024
This is a demonstration of the proposed MAP function used to solve a challenge provided by Oz_Sunray and Wyn Hopkins on data cleansing and vote counting. The key element of my solution is to build a table of validated votes for each voter ID and each candidate (i.e. an array of arrays - much the same as any other problem of interest). The formula is modularised using Lambda functions. They are not that pretty, but neither are they impossible to follow.
"Worksheet formula" = AggregateVotesλ( tblVOTES[VoterID], tblVOTES[[Vote1]:[Vote3]], weight, tblAvailable[AVAILABLE] )
which calls
/* "FUNCTION NAME: AggregateVotesλ" "DESCRIPTION: Aggregates all votes to candidate colors" */ /* "REVISIONS: Date Developer Description" " 08 Sep 2024 Peter Bartholomew Original Development" */ AggregateVotesλ = LAMBDA(voterID, votesCast, weight, available, LET( person, TOROW(UNIQUE(voterID)), votes, MAPλ(person, AssignVotesλ(voterID, votesCast, weight, available)), count, HSTACK(available, BYROW(votes, SUM)), return, TAKE(SORT(count, 2, -1), 4), return ) );
to build and aggregate the table of votes. This, in turn, calls another Lambda function to clean and validate the votes corresponding to a single voterID
/* "FUNCTION NAME: AssignVotesλ" "DESCRIPTION: Validates and assigns votes to candidate colors for a single voterID " */ /* "REVISIONS: Date Developer Description" " 08 Sep 2024 Peter Bartholomew Original Development" */ AssignVotesλ = LAMBDA(voterID, ballots, weight, available, LAMBDA(p, LET( // "In the case of duplicate ballots return first" votesCast, XLOOKUP(p, voterID, ballots), // "Remove invalid and repeated votes" valid, COUNTIFS(available, votesCast), distinct, EXPAND(UNIQUE(FILTER(votesCast, valid), TRUE), , 3, ""), // "Assign votes to candidate colours" assigned, BYROW((distinct = available) * weight, SUM), assigned ) ) );
I took the problem from Wyn's recording
https://www.youtube.com/watch?v=FIcxyLqzWcE
I have posted a further application of the MAPλ using the OP's (joelb95) workbook. In the workbook I have applied the home-rolled helper function both to the task of maintaining multiple running totals and to a FIFO calculation of the same 'Lots'.
The FIFO calculation filters the event table by each 'Lot' in turn but MAPλ stack the tables as an array.
As an alternative to the array of tables the user can select an array of grouped tables that focus on the outputs along with associated input costs.
A key element of both MAPλ and the FIFOλ functions is that they are much easier to use than they were to write. But then, that is the whole idea of using modern methods and LAMBDA in particular!