Can change shape of range with ARRAYFORMULA() in Google Sheets? - google-sheets

My intention is to convert a single line of data into rows consist of a specific number of columns in Google Sheets.
For example, starting with the raw data:
A
B
C
D
E
F
1
id1
attr1-1
attr2-1
id2
attr2-1
attr2-2
And the expected result is:
(by dividing columns by three)
A
B
C
1
id1
attr1-1
attr1-2
2
id2
attr2-1
attr2-2
I already know that it's possible a bit manually, like:
=ARRAYFORMULA({A1:C1;D1:F1})
But I have to start over with it every time the target range is moved OR the subset size needs to be changed (in the case above it was three)!
So I guess there will be a much more graceful way (i.e. formula does not require manual update) to do the same thing and suspect ARRAYFORMULA() is the key.
Any help will be appreciated!

I added a new sheet ("Erik Help") where I reduced your manually entered parameters from two to one (leaving only # of columns to be entered in A2).
The formula that reshapes the grid:
=ArrayFormula(IFERROR(VLOOKUP(SEQUENCE(ROUNDUP(COUNTA(7:7)/A2),A2),{SEQUENCE(COUNTA(7:7),1),FLATTEN(FILTER(7:7,7:7<>""))},2,FALSE)))
SEQUENCE is used to shape the grid according to whatever is entered in A2. Rows would be the count of items in Row 7 divided by the number in A2 (rounded to the nearest whole number); and the columns would just be whatever number is entered in A2.
Example: If there are 11 items in Row 7 and you want 4 columns, ROUNDUP(11/4)=3 rows to the SEQUENCE and your requested 4 columns.
Then, each of those numbers in the grid is VLOOKUP'ed in a virtual array consisting of a vertical SEQUENCE of ordered numbers matching the number of data pieces in Row 7 (in Column 1) and a FLATTENed (vertical) version of the Row-7 data pieces themselves (in Column 2). Matches are filled into the original SEQUENCE grid, while non-matches are left blank by IFERROR

Though it's a bit messy, managed to get it done thanks to SEQUENCE() function anyway.
It constructs a grid by accepting number of rows/columns input, and that was exactly I was looking for.
For reference set up a sheet with the sample data here:
https://docs.google.com/spreadsheets/d/1p972tYlsPvC6nM39qLNjYRZZWGZYsUnGaA7kXyfJ8F4/edit#gid=0

Use a custom formula
Although you already solved this. If you are doing this kind of thing a lot, it could be beneficial to look into Apps Script and custom formulas.
In this case you could use something like:
function transposeSingleRow(range, size) {
// initialize new range
let newRange = []
// initialize counter to keep track
let count = 0;
// start while loop to go through row (range[0])
while (count < range[0].length){
// add a slice of the original range to the new range
newRange.push(
range[0].slice(count, count + size)
);
// increment counter
count += size;
}
return newRange;
}
Which works like this:
The nice thing about the formula here is that you select the range, and then you put in a number to represent its throw, or how many elements make up a complete row. So if instead of 3 attributes you had 4, instead of calling:
=transposeSingleRow(A7:L7, 3)
you could do:
=transposeSingleRow(A7:L7, 4)
Additionally, if you want this conversion to be permanent and not dependent on formula recalculation. Making it in run fully in Apps Script without using formulas would be neccesary.
Reference
Apps Script
Custom Functions

Related

ArrayFormula to calculate previous rows

I have 3 sheets:
Sheet1 - list of transactions (Account, Credit, Debit, Date)
Sheet2 - list of transactions (Account, Credit, Debit, Date)
Sheet3 (I plan to lock it) - combined list of transactions, sorted by Date
Sheet3 looks like:
I need to add 1 more column to Sheet3 to count current balance for certain row to be like:
I'm able to do this with formula:
=SUM(FILTER($B$2:$B$8, ROW($A$2:$A$8) <= ROW($A2), A$2:A$8=$A2)) - SUM(FILTER($C$2:$C$8, ROW($A$2:$A$8) <= ROW($A2), A$2:A$8=$A2))
But this one I need continuously drag down.
Question: Is there way convert this formula to ArrayFormula, to avoid dragging
In G2 on sheet 3 I entered
=ArrayFormula(if(A2:A="",,mmult((A2:A=transpose(A2:A))*(row(A2:A)>= TRANSPOSE(row(A2:A)))*(transpose(B2:B)-transpose(C2:C)),row(A2:A)^0)))
See if that works for you?
In Sheet3 row 1, put your headers.
In Sheet3!A2, put
=sort({filter(Sheet1!A2:D,not(isblank(Sheet1!A2:A)));filter(Sheet2!A2:D,not(isblank(Sheet2!A2:A))),4,true)
In Sheet3!E2, put
=mmult(transpose(arrayformula(arrayformula(array_constrain(A2:A,counta(A2:A),1)=transpose(array_constrain(A2:A,counta(A2:A),1)))
*arrayformula(array_constrain(row(A2:A),counta(A2:A),1)<=transpose(array_constrain(row(A2:A),counta(A2:A),1))))),
arrayformula(array_constrain(B2:B,counta(A2:A),1)-array_constrain(C2:C,counta(A2:A),1))
To see why, let's temporarily remove the array_constrain(...,counta(...),1) wrappings, which is meant to auto detect the last data row:
=mmult(transpose(arrayformula(arrayformula(A2:A9=transpose(A2:A9))
*arrayformula(row(A2:A9)<=transpose(row(A2:A9))))),
arrayformula(B2:B9-C2:C9))
arrayformula(B2:B9-C2:C9) are the running sums of column B - column C (ie. credit - debit). It is a column vector with the length of your data size.
We want to, for each row, 1) filter this vector by comparison to column A (ie. account name) & 2) filter this vector by whether the running sums are below or above the row in question.
arrayformula(A2:A9=transpose(A2:A9)) does 1). arrayformula(row(A2:A9)<=transpose(row(A2:A9))) does 2).
We want elementwise product between the 2 matrices in order to compose the filter. Hence, arrayformula(...*...).
The columns of our filters are meant to be applied to the running sums. To use matrix multiplication, we can keep the column vector of running sums as the post-multiplier; and transpose the filter matrix as pre-multiplier so that the rows of the transposed matrix are multiplied (ie. applied) to the running sums. Hence, mmult(transpose(...),...).
Add back the array_constrain trick. And we are done.
Feel free to experiment with alternate placings of arrayformula. But remember to keep the () brackets wherever you omit arrayformula. Example:
=arrayformula(mmult(transpose(((array_constrain(A2:A,counta(A2:A),1)=transpose(array_constrain(A2:A,counta(A2:A),1)))
*(array_constrain(row(A2:A),counta(A2:A),1)<=transpose(array_constrain(row(A2:A),counta(A2:A),1))))),
(array_constrain(B2:B,counta(A2:A),1)-array_constrain(C2:C,counta(A2:A),1))))
Nonetheless, the 1 formula solution is computationally inefficient compared to individually spread formula per cell. That is because, without mutating the formula per row, we are forced to compute the filters as full n-by-n matrices where n is your data size.
Whereas, if in E2 we put =sum(filter(B$2:B2-C$2:C2,A$2:A2=A2)) and spread to the end by double right-clicking the square on bottom right when you select E2, the formula mutates per row, saving the row index comparison entirely, and also cutting the comparison to column A logarithmically.
Granted, we probably shouldn't rely on Google Sheet for a large database (e.g. >100k entries). But even for thousands of entries, if you square the amount of computations required, getting the results in browser becomes impractically slow well before one may expect.

Automatically fill sum column

I have a table of items with its buying and selling rates. Against each transaction, I wanted to show the item qty currently available and the current cost. Here is a screenshot of the table.
For the columns of Qty and Rate, I have used the formulas as shown below:
=SUMPRODUCT(C$2:C-F$2:F,B$2:B=B2,A$2:A<=A2)
=SUMPRODUCT(I$2:I/J2,B$2:B=B2,A$2:A<=A2)
Is it possible to convert SUMPRODUCT formulas as an array formula (returning array) so that it automatically fills whenever a row is added? Something like filling a column using FILTER or QUERY?
Regards,
Pravin Kumar.
I am making an assumption that this is your goal:
To automatically fill the quantity remaining column with the result of the difference between the two values, if and only if, there are two values provided.
A solution to this objective is:
=arrayformula(if(and(C2:C<>””, F2:F<>””), C2:C-F2:F, “”))
This produces a formula that subtracts C from F only if both C and F have values, and for all rows that have values in both C and F. If one of the rows does not have a value, the result will yield “” (blank). This formula should be posted at the top of the column where you want these results to start. In your screenshot example this would be cell J2. NOTE: 0 is still a value, and also that an array formula will not overwrite manually input data, so once you paste that function in J2, you will have to clear the cells below in order for it to auto populate.

Using COUNTIFS in an array formula

I'm trying to count the number of times something of a given type occurs and I need this behaviour to automatically expand to inserted rows. Something like:
=Arrayformula(COUNTIFS(I:I,I:I,H:H,H:H,G:G,G:G))
The nested countif formula will result in a correct value when used on a single row but currently the array formula is outputting 1 all the way down.
My data resembles:
Column1 Column2 Column3 Result
--------------------------------------------
apple green eaten x
orange orange noteaten x
apple red eaten x
orange orange noteaten x
apple green eaten x
...
The x column is where the arrayformula would output.
X on Row 1 should look through all the data and count up the number of green apples eaten, the next row would count noneaten orange oranges, and so on. I know that arrayformula doesn't take aggregate functions but I didn't find anything on alternatives to countif.
Unfortunately, in Google Sheets, COUNTIFS can not be iterated over an array, as eg COUNTIF can (at the time of writing this, anyway).
You would need to resort to MMULT, something like:
=ArrayFormula(IF(ROW(G:G)=1,"Result",MMULT((G:G=TRANSPOSE(G:G))*(H:H=TRANSPOSE(H:H))*(I:I=TRANSPOSE(I:I)),SIGN(ROW(G:G)))))
but be aware there appears to be a limitation in Sheets whereby the 2D array formed by G:G=TRANSPOSE(G:G) etc cannot exceed 10 million elements. This corresponds to a maximum of 3162 rows.
Another option is to use concatenation of strings:
=ArrayFormula(COUNTIF(G:G&CHAR(9)&H:H&CHAR(9)&I:I,G:G&CHAR(9)&H:H&CHAR(9)&I:I))
which gets around the "3162" limitation. CHAR(9) is a tab character, but it could be any character that you are certain will not appear in your data.
This is old but I found two workarounds.
Adding an if statement to the ArrayFormula does the trick.
E.g.:
=ArrayFormula(IF(I:I="","",COUNTIFS(I:I,I:I,H:H,H:H,G:G,G:G)))
Otherwise, if you can afford to have an extra column, you could combine the data in your three columns and run a COUNITF on that.
That said, you'd need to include an if statement in that to exclude the count for empty rows.
Column K:
=ArrayFormula(G:G&H:H&I:I)
In the Result Column:
=ArrayFormula(IF(I:I="","",COUNTIF(K:K,K:K)))

Google Spreadsheet sum which always ends on the cell above

How to create a Google Spreadsheet sum() which always ends on the cell above, even when new cells are added? I have several such calculations to make on each single column so solutions like this won't help.
Example:
On column B, I have several dynamic ranges which has to be summed. B1..B9 should be summed on B10, and B11..B19 should be summed on B20. I have tens such calculations to make. Every now and then, I add rows below the last summed row , and I want them to be added to the sum. I add a new row (call it 9.1) before row 10, and a new raw (let's call it 19.1) before row 20. I want B10 to contain the sum of B1 through B9.1 and B20 to contain the sum of B11:B19.1.
On excel, I have the offset function which does it like charm. But how to do it with google spreadsheet? I tried to use formulas like this:
=SUM(B1:INDIRECT(address(row()-1,column(),false))) # Formula on B10
=SUM(B11:INDIRECT(address(row()-1,column(),false))) # Formula on B20
But on Google Spreadsheet, all it gives is a #name error.
I wasted hours trying to find a solution, maybe someone can calp?
Please advise
Amnon
You are probably looking for formula like:
=SUM(INDIRECT("B1:"&ADDRESS(ROW()-1,COLUMN(),4)))
Google Spreadsheet INDIRECT returns reference to a cell or area, while - from what I recall - Excel INDIRECT returns always reference to a cell.
Given Google's INDIRECT indeed has some hard time when you try to use it inside SUM as cell reference, what you want is to feed SUM with whole range to be summed up in e.g. a1 notation: "B1:BX".
You get the address you want in the same way as in EXCEL (note "4" here for row/column relative, by default Google INDIRECT returns absolute):
ADDRESS(ROW()-1,COLUMN(),4)
and than use it to prepare range string for SUM function by concatenating with starting cell.
"B1:"&
and wrap it up with INDIRECT, which will return area to be sum up.
REFERRING TO BELOW ANSWER from Druvision (I cant comment yet, I didn't want to multiply answers)
Instead of time consuming formulas corrections each time row is inserted/deleted to make all look like:
=SUM(INDIRECT(ADDRESS(ROW()-9,COLUMN(),4)&":"&ADDRESS(ROW()-1,COLUMN(),4)))
You can spare one column in separate sheet for holding variables (let's name it "def"), let's say Z, to define starting points e.g.
in Z1 write "B1"
in Z2 write "B11"
etc.
and than use it as variable in your sum by using INDEX:
SUM(INDIRECT(INDEX(def!Z:Z,1,1)&":"&ADDRESS(ROW()-1,COLUMN(),4))) - sums from B1 to calculated row, since in Z1 we have "B1" ( the 1,1 in INDEX(...,1,1) )
SUM(INDIRECT(INDEX(def!Z:Z,2,1)&":"&ADDRESS(ROW()-1,COLUMN(),4))) - sums from B11 to calculated row, since in Z2 we have "B11" ( the 2,1 in INDEX(...,2,1) )
please note:
Separate sheet named 'def' - you don't want row insert/delete influence that data, thus keep it on side. Useful for adding some validation lists, other stuff you need in your formulas.
"Z:Z" notation - whole column. You said you had a lot of such formulas ;)
Thus you preserve flexibility of defining starting cell for each of your formulas, which is not influenced by calculation sheet changes.
By the way, wouldn't it be easier to write custom function/script summing up all rows above cell? If you feel like javascripting, from what I recall, google spreadsheet has now nice script editor. You can make a function called e.g. sumRowsAboveMe() and than just use it in your sheet like =sumRowsAboveMe() in sheet cell.
Note: you might have to replace commas by semicolons
NOTE
After testing this answer, it will only work if the sum is in a different column due to a circular dependency error. Otherwise, the solution is valid.
It's a bit of algebra, but we can take advantage of Spreadsheets' lower right corner drag.
=SUM(X:X) - SUM(X2:X)
Where X is the column you are working with and X2 is your ending point. Drag the formula down and Sheets will increment the X2, thus changing the ending point.
*You mentioned that you had tens of such calculations to make. So in order to fit your exact need, we would subtract your last summation to get that "middle" range that we wanted.
e.g.
B1..B9 should be summed on B10, and B11..B19 should be summed on B20
Because of the circular dependency error mentioned earlier, I can't solve it exactly and put the sum on the same line, but this could work in other cases where the sum needs to be stored in a different column.
=SUM(B:B) - SUM(B9:B) //Formula on C10 (Sum of B1..B9)
=SUM(B:B) - SUM(B19:B) - B10 // Formula on C20 (Sum of B11..B19)
This is based on #PsychoFish, here is the solution:
=SUM(INDIRECT(SUBSTITUTE(ADDRESS(1,COLUMN(),4),"1","")&"3:"&ADDRESS(ROW()-1,COLUMN(),4)))
Simply replace the "3:" for the row to start sum.
#PsychoFish is correct but cannot be dragged and copied since the column is literal and hard coded, and #Druvision was in the right direction but was wrong... basically ended up with the same issue of having to re-enter the ranges and then sliding the formulas over and over.
You guys are making this harder than you have to. I just leave a couple of empty rows above by "sum" row (you can format them to be filled with color or something to keep them from being inadvertently used), then just add your new rows just above those special rows.
Agree with what user7255446 said that everyone is overcomplicating. Keep one row blank before your sum row. And then whenever you want to insert a new row, click on your blank row and use "Insert row ABOVE" instead of "insert row below". Your sum formula will automatically adjust.
Example: I want to sum from B1 to B19. I leave row 20 blank. In cell B21, put =SUM(B1:B20). Then if you ever need to insert a new row, click on row 20 and choose "Insert row above". The sum formula automatically changes to =SUM(B1:B21) for you. And of course your sum cell is now B22.
General syntax:
=SUM(INDIRECT(cell_reference_as_string1 &":"& cell_reference_as_string2)
with for example:
cell_reference_as_string1 = ADDRESS(ROW(),COLUMN(),4)
cell_reference_as_string2 = ADDRESS(ROW()-1,COLUMN(),4)
I like how #abernier describes the general solution. So far only alphabet-based A1 notation (A being first column, 1 being first row) are being used. It keeps confusing me, especially when thinking of number of columns left of another column. I like the number-based R1C1 notation much better. To use R1C1 notation for INDIRECT, you need to pass FALSE like so:
=SUM(INDIRECT("R1C"&COLUMN()&":R"&(ROW()-1)&"C"&COLUMN(), FALSE))
I hope you find that helpful, too.
OFFSET() can be used/abused for this purpose. Give it the absolute address of the top left of the range, 0 and 0 for the row/column offsets, and the height/width of the range. Let OFFSET() be the argument to SUM(), SUMIF(), etc.
ROW() and COLUMN() are handy when computing the desired height/width. Be sure to remember to subtract one to exclude the current row/column, or else you're liable to end up with a circular reference. If you have header rows/columns, subtract for them too.
For example, to sum everything from A2 down, excluding the current row, try:
=SUM(OFFSET($A$2,0,0,ROW()-2,1))
To sum everything to the left of the current cell, wherever it may be, try:
=SUM(OFFSET(INDIRECT("RC1",FALSE),0,0,1,COLUMN()-1))
Now let's flip things upside down, to show that this works in the other direction. Suppose you want to sum the B column, starting below the current row, until (and including) row #10. Try this:
=SUM(OFFSET($B$10,ROW()-9,0,10-ROW(),1))
You can avoid negative offsets, while still summing column B:
=SUM(OFFSET(INDIRECT("RC2",FALSE),1,0,10-ROW(),1))
Remove the "2" to instead sum the current column:
=SUM(OFFSET(INDIRECT("RC",FALSE),1,0,10-ROW(),1))
(Credit to Tom Sharpe, who commented above.) INDEX() can be used in a range expression. You might prefer this over OFFSET(), so I'm putting it here. The following sums everything from G1 down to the row above the current:
=SUM(G1:INDEX(G:G,ROW()-1))
Here's how I do it.
This formula does not require you to edit or enter anything about the particular column you would like to sum
=SUM(INDIRECT(CONCATENATE(address(1,column(),4),":",LEFT(address(1,column(),4),1))&ROW()-1))
The answer by #PsychoFish led me in the correct way.
The only issue that I had to rewrite the formula again from each column and each sum. So here is the improved formula, which sums the previous 9 cells on the same column, without hardcoding the column or row numbers:
=SUM(INDIRECT(ADDRESS(ROW()-9,COLUMN(),4)&":"&ADDRESS(ROW()-1,COLUMN(),4)))
The only issue is that I had to rewrite the formulas if someone adds or deletes a row. In this case I should change 9 to 10 or 8 corrspondingly.

How can I generate a random number (or Cell) within a given range? Witout using the same result twice?

I am trying to generate a random Cell from specific range:
I need to each cell Row to generate a random selection from a a specific column (range)
below is a picture of my set up and my failed attempts:
You can do this like this:
Add a random number next to your data set using =RAND(). I've used column B, but you can put it wherever you like.
Add this formula to cells C2 to H2
=INDEX($A$2:$A$21,RANK.EQ(INDEX($B$2:$B$21,COLUMN()-2),$B$2:$B$21))
How it works:
RAND() returns a random number in the range [0..1) This is used as a random sort order for your data
Breaking down the formula:
COLUMN()-2 returns a sequential number 1..6 for columns C to H
INDEX($B$2:$B$21, ... ) returns the 1st to 6th number from the random number list
RANK.EQ( ... ,$B$2:$B$21) returns the position of the random number in the sorted random number list, 1..20.
=INDEX($A$2:$A$21, ... ) returns an item from your data set, based on the random rank from above.
Note: This will return a new randon sample each time Excel recalculates.
The only way to make a random selection that does not repeat is to make an array of integers, than randomize it, and than take out one by one.
For example you start with:
1 2 3 4 5 6 7
you randomize it (swap elements randomly)
2 4 5 1 7 3 6
than you take elements out one by one. (keep an index of how much elements you used in some cell)
Ok here is one trick, add another column with random numbers, than select both random column and range column and click the sort button (picture is from libreoffice but there is a similar button in excel) This will randomize your range column. Than you simply assign values to "w-1" : "w-6" like this =B2, =B3, =B4, =B5, =B6, =B7
I tried this and it works like you wanted, the only problem is it will shuffle values in your range column.

Resources