I have a series of cells with variations of a long formula which involves filters.
It takes a certain array, does some counting to some of its columns (checking the number of nonempty cells, then the number of positive cells), but then I need to filter that array further based on the sum of certain of its columns. The important nuance is that I need this formula to be pretty uniform, only changing the first two variables (label and countedColumns) for ease of mass implementation. This seemed to work great until I had to reference the proper cells inside the row for the sake of filtering, and to sum the values of those cells.
So far I tried something roughly like this (failing at the filtering stage):
=LET(
label, " SomeLabel ",
countedColumns, {30,33},
array, IF(ArrayOfTheDay!$A$3:$ZZ$999="","",ArrayOfTheDay!$A$3:$ZZ$999),
countedCells, CHOOSECOLS(array, countedColumns),
countedCellsClean, IF(countedCells="","",countedCells),
attemptCount, SUMPRODUCT(--(countedCellsClean<>"")),
successCount, SUM(countedCellsClean),
filteredSuccessData, FILTER(array, SUM(CHOOSECOLS(array,countedColumns)>0)), // this is not working
listOfSuccessDataAsText, // this will TextJoin/Concatenate certain cells from List
successCountText, IF(successCount>0, successCount&": ", "-"),
attemptCountText, IF(attemptsCount>0, " attempts: "&attemptCount, ""),
CONCATENATE(label, successCountText, ": ", listOfSuccessDataAsText," separator ", attemptCountText)
)
How can I properly choose the right columns within a row for the sake of filtering rows based on their sums, or how else can I approach it so as to not need to specify the columns multiple times per such formula? (I.e. I would rather not have to manually write a bunch of (column1address>0)+(column2address>0)+() entries in each and every instance of this filter, and it would be best not to use addresses at all.)
The second argument of FILTER
expects an array with the same number of rows as the first argument. However, SUM
returns a single scalar value, not an array.
If you want to turn multiple columns into a single column containing their sum, then use the MMULT
function
summedColumn, MMULT(countedCells,--TRANSPOSE(INDEX(countedCells,1,)=INDEX(countedCells,1,))),
filteredSuccessData, FILTER(array, summedColumn>0)
or the ByRow
function:
summedColumn, ByRow(countedCells, LAMBDA(rw, SUM(rw))),
filteredSuccessData, FILTER(array, summedColumn>0)