I am a song leader at our church, and I am using a spreadsheet to track how recently we have used each hymn in our hymnbook. (In an effort to give some rotation to the songs we sing each week.)
Using some very helpful guides on Exceljet, I was able to construct a formula that does almost exactly what I want... It tells me how many weeks ago we used each song in our hymnbook. The problem is that it currently only shows me the FIRST time we used the song, not the LAST or most recent time, which is what I am really wanting to do.
The key issue I am stuck on is how to get the last entry from a MMULT() formula. A simplified example is shown below... I want to return the last date from column A where example hymn #15 is listed in columns B, C, or D.
My current formula does a great job of getting the first matching row (Row 3) but does not give me the last row (Row 5).
{=INDEX(ServiceDate,MATCH(1,MMULT(--(HymnNumbers=F2),TRANSPOSE(COLUMN(HymnNumbers)^0)),0))}
(ServiceDate= A:A and HymnNumbers = B:D)
Yes, I realize that there are other ways to find the last match, with MATCH and LOOKUP if I build a very complicated expression that searches for the last match in each column individually and finds the largest date value, but performance is also an issue with the size of the data in the actual spreadsheet I am using.
Is there is something simple that I can tweak in my current MMULT function to return the last match instead of the first one?
If dates are sorted you can reduce your formula to
=INDEX(A:A,AGGREGATE(14,6,ROW(A:A)/(B:D=F2),1))
If you have Office365 and dynamic formulas then you can use simply below formula.
=MAX(FILTER(A2:A9,(B2:B9=F2)+(C2:C9=F2)+(D2:D9=F2)))
If you do not have Office365 then try below Array formula.
=MAX(IF(B2:B9=F2,A2:A9,""),IF(C2:C9=F2,A2:A9,""),IF(D2:D9=F2,A2:A9,""))
Press CTRL+SHIFT+ENTER to evaluate the formula as it is an array formula.
Here's a longer formula that will return the last match. I'm not sure if it fits your definition of "very complicated". This version is longer since it does not use the latest version of excel.
=INDEX(A1:A10,MATCH(LARGE(MMULT(--(B1:D10=F2),TRANSPOSE(COLUMN(B1:D10)^0))*ROW(A1:A10),1),MMULT(--(B1:D10=F2),TRANSPOSE(COLUMN(B1:D10)^0))*ROW(A1:A10),0))
Your formula can be amended as follows...
=INDEX(ServiceDate,MATCH(2,1/MMULT(--(HymnNumbers=F2),TRANSPOSE(COLUMN(HymnNumbers))^0)))
However, you can avoid using MMULT as follows...
=INDEX(ServiceDate,LARGE(IF(HymnNumbers=F2,ROW(ServiceDate)-MIN(ROW(ServiceDate))+1),1))
Note that both these formulas need to be confirmed with CONTROL+SHIFT+ENTER.
This question was asked and as the answer to the specific question was a typo it was deleted:
https://stackoverflow.com/questions/59289065/excel-non-adjecent-cells-as-input-to-array-function-min-and-isblank
Here is the question:
I'm trying to find the minimum of two (non-adjacent) cells per column and sum these for a number of columns (13 in total).
What complicates it is that I'd like the function to treat empty cells as zero.
I can get it to work as long as the cells are adjacent, but when they are not, excel gives a "too many arguments for this function" pop-up.
The formula I have for adjecent cells is this (not exactly pretty, sorry!):
{=SUM(MIN(IF(ISBLANK(P3:P4);0;P3:P4));MIN(IF(ISBLANK(Q3:Q4);0;Q3:Q4));MIN(IF(ISBLANK(R3:R4);0;R3:R4));MIN(IF(ISBLANK(S3:S4);0;S3:S4));MIN(IF(ISBLANK(T3:T4);0;T3:T4));MIN(IF(ISBLANK(U3:U4);0;U3:U4));MIN(IF(ISBLANK(V3:V4);0;V3:V4));MIN(IF(ISBLANK(W3:W4);0;W3:W4));MIN(IF(ISBLANK(X3:X4);0;X3:X4));MIN(IF(ISBLANK(Y3:Y4);0;Y3:Y4));MIN(IF(ISBLANK(Z3:Z4);0;Z3:Z4));MIN(IF(ISBLANK(AA3:AA4);0;AA3:AA4));MIN(IF(ISBLANK(AB3:AB4);0;AB3:AB4)))}
This gives the desired output in the column "person months total".
I have tried to use the CHOOSE function for non-adjacent cells as a test (similar to this question), but this gives the "There's something wrong with this formula" pop-up
=SUM(MIN(IF(ISBLANK(CHOOSE{1;2};P16;P18));0;CHOOSE({1;2};P16;P18)))
So now I'm wondering, can this be done at all? Am I missing something?
I would appreciate the help!
Kind regards,
Amy
While the answer to this question was that there was a missing ( after the first CHOOSE, I started working on a simpler version to that hideous long formula and wanted to post it here. So the question is, "Is there a simpler method not using vba?"
With the ranges actually being adjacent by rows, but comparing column by column we can use MMULT in an array form. By using MMULT we can create an array of the smallest numbers and 0s and sum them:
=SUM(MMULT(N(IF(A1:E1>A2:E2,IF(A2:E2<>"",A2:E2),IF(A1:E1<>"",A1:E1))),TRANSPOSE(COLUMN(A1:E1)^0)))
This is an array formula and must be confirmed with Ctrl-Shift-Enter instead of Enter when exiting edit mode.
EDIT: way over thought it, this is much simpler:
=SUM(IF(A1:E1>A2:E2,IF(A2:E2<>"",A2:E2),IF(A1:E1<>"",A1:E1)))
Still an array formula.
I'm using counting invoice numbers (text) in a table's column, but the Excel formula seems to be confusing some values.
I copied small sample of these - please refer to below:
The formulas are as follow:
=COUNTIFS(A1:A19,A1)
=COUNTIF(A1:A19,A1)
As you can see these invoice numbers differ and the results of these functions suggest as if all were the same.
I googled it for 1 hour but I didn't find such as issue as mine.
If anybody had any clue why could this behave in such way I'll be super grateful!
Rob
Each time you copy down this formula it will add 1 row to each. For example the second row of datas formula will be =COUNTIFS(A2:A20,A2). To lock these cells in the formula use $
Your formula should be =COUNTIFS(A$1:A$19,A1)
I've solved this myself:
ROOTCAUSE
Excel tried to be helpful and read these invoice numbers as actual numbers (despite these being defined already in Power Query as text)
Then, Excel fooled me and despite showing that it works on it as a string (I was evaluating the formula) it worked on it as number
Above means that it transformed exemplary "00100001010000018525" to 1.00001E+17, which cut down this to "100001010000018000" - that's the moment Excel stopped fooling around and showed that value in the formula bar.
I think I don't need to tell why countif perceived all these values as equal.
SOLUTION
I simply appended one letter after each invoice number to get e.g. "00100001010000018525a" what forces Excel to quit its gimmicks and games.
Case closed.
I suspect this is a bug in COUNTIF, or maybe by design.
However, to workaround this in the formula, without having to change your data, try adding a wild-card character:
=COUNTIF(A1:A19,"*"&A1)
I have been working on this issue for over two hours and I am beginning to question my sanity. I have used vlookup many times in the past, but now it is just not working as expected. I am trying to replicate a LEFT-JOIN (from SQL) in Excel using vlookup.
Here is the formula I am using on cell G2:
Here is the table of all the values I want to lookup values for.
Here is the table I am using as a reference:
I am only searching one column to simplify the example. Cell G2 contains the formula which is applied to the cells under it as well. As you can see from the first image, not only is it not matching but there is some pretty weird behavior going on.
I have removed duplicates for both tables.
I have unformatted the data to plain text values
I have tried this formula on three different computers
Regardless, I keep kept getting the same result! I am starting to lose sanity.
Does anyone have any idea?
Thank you
If you want exact matches, you should be using FALSE as the last parameter
The lookup table must be sorted in ascending order. Cell A369 appears to have a value lower than the row before it, A368. There are several instances of "lower" values occurring after a higher value.
Before you go insane, consider sorting the range $A$368 thru $A$679, and see if that makes a difference.
Otherwise, time to ditch VLOOKUP, and use instead INDEX and MATCH.
Sometimes you can get the #N/A error if your lookup_value isn't "clean". If that's the case, try this formula:
=VLOOKUP(TRIM(CLEAN(A2)),$A$368:$A$697,1,FALSE)
Additionally, your table_array may have "unclean" data, so you'll need to scrub that first before you're able to find a match. To do that, use this array formula, committing it with Ctrl + Shift + Enter:
=VLOOKUP(A2,TRIM(CLEAN($A$368:$A$697)),1,FALSE)
Have you tried doing a MATCH? does it need to be VLOOKUP?
if you are happy using match try:
=IF(MATCH(A2,$A$368:$A$697,0),A2,"NO MATCH")
Does anyone know the formula to find the value of the last non-empty cell in a column, in Microsoft Excel?
Using following simple formula is much faster
=LOOKUP(2,1/(A:A<>""),A:A)
For Excel 2003:
=LOOKUP(2,1/(A1:A65535<>""),A1:A65535)
It gives you following advantages:
it's not array formula
it's not volatile formula
Explanation:
(A:A<>"") returns array {TRUE,TRUE,..,FALSE,..}
1/(A:A<>"") modifies this array to {1,1,..,#DIV/0!,..}.
Since LOOKUP expects sorted array in ascending order, and taking into account that if the LOOKUP function can not find an exact match, it chooses the largest value in the lookup_range (in our case {1,1,..,#DIV/0!,..}) that is less than or equal to the value (in our case 2), formula finds last 1 in array and returns corresponding value from result_range (third parameter - A:A).
Also little note - above formula doesn't take into account cells with errors (you can see it only if last non empty cell has error). If you want to take them into account, use:
=LOOKUP(2,1/(NOT(ISBLANK(A:A))),A:A)
image below shows the difference:
This works with both text and numbers and doesn't care if there are blank cells, i.e., it will return the last non-blank cell.
It needs to be array-entered, meaning that you press Ctrl-Shift-Enter after you type or paste it in. The below is for column A:
=INDEX(A:A,MAX((A:A<>"")*(ROW(A:A))))
Here is another option: =OFFSET($A$1;COUNTA(A:A)-1;0)
I know this question is old, but I'm not satisfied with the answers provided.
LOOKUP, VLOOKUP and HLOOKUP has performance issues and should really never be used.
Array functions has a lot of overhead and can also have performance issues, so it should only be used as a last resort.
COUNT and COUNTA run into problems if the data is not contiguously non-blank, i.e. you have blank spaces and then data again in the range in question
INDIRECT is volatile so it should only be used as a last resort
OFFSET is volatile so it should only be used as a last resort
any references to the last row or column possible (the 65536th row in Excel 2003, for instance) is not robust and results in extra overhead
This is what I use
when the data type is mixed: =max(MATCH(1E+306,[RANGE],1),MATCH("*",[RANGE],-1))
when it's known that the data contains only numbers: =MATCH(1E+306,[RANGE],1)
when it's known that the data contains only text: =MATCH("*",[RANGE],-1)
MATCH has the lowest overhead and is non-volatile, so if you're working with lots of data this is the best to use.
Inspired by the great lead given by Doug Glancy's answer, I came up with a way to do the same thing without the need of an array-formula. Do not ask me why, but I am keen to avoid the use of array formulae if at all possible (not for any particular reason, it's just my style).
Here it is:
=SUMPRODUCT(MAX(($A:$A<>"")*(ROW(A:A))))
For finding the last non-empty row using Column A as the reference column
=SUMPRODUCT(MAX(($1:$1<>"")*(COLUMN(1:1))))
For finding the last non-empty column using row 1 as the reference row
This can be further utilized in conjunction with the index function to efficiently define dynamic named ranges, but this is something for another post as this is not related to the immediate question addressed herein.
I've tested the above methods with Excel 2010, both "natively" and in "Compatibility Mode" (for older versions of Excel) and they work. Again, with these you do not need to do any of the Ctrl+Shift+Enter. By leveraging the way sumproduct works in Excel we can get our arms around the need to carry array-operations but we do it without an array-formula. I hope someone out there may appreciate the beauty, simplicity and elegance of these proposed sumproduct solutions as much as I do. I do not attest to the memory-efficiency of the above solutions though. Just that they are simple, look beautiful, help the intended purpose and are flexible enough to extend their use to other purposes :)
Hope this helps!
All the best!
This works in Excel 2003 (& later with minor edit, see below). Press Ctrl+Shift+Enter (not just Enter) to enter this as an array formula.
=IF(ISBLANK(A65536),INDEX(A1:A65535,MAX((A1:A65535<>"")*(ROW(A1:A65535)))),A65536)
Be aware that Excel 2003 is unable to apply an array formula to an entire column. Doing so yields #NUM!; unpredictable results may occur! (EDIT: Conflicting information from Microsoft: The same may or may not be true about Excel 2007; problem may have been fixed in 2010.)
That's why I apply the array formula to range A1:A65535 and give special treatment to the last cell, which is A65536 in Excel 2003. Can't just say A:A or even A1:A65536 as the latter automatically reverts to A:A.
If you're absolutely sure A65536 is blank, then you can skip the IF part:
=INDEX(A1:A65535,MAX((A1:A65535<>"")*(ROW(A1:A65535))))
Note that if you're using Excel 2007 or 2010, the last row number is 1048576 not 65536, so adjust the above as appropriate.
If there are no blank cells in the middle of your data, then I would just use the simpler formula, =INDEX(A:A,COUNTA(A:A)).
An alternative solution without array formulas, possibly more robust than that of a previous answer with a (hint to a) solution without array formulas, is
=INDEX(A:A,INDEX(MAX(($A:$A<>"")*(ROW(A:A))),0))
See this answer as an example.
Kudos to Brad and barry houdini, who helped solving this question.
Possible reasons for preferring a non-array formula are given in:
An official Microsoft page (look for "Disadvantages of using array formulas").
Array formulas can seem magical, but they also have some disadvantages:
You may occasionally forget to press CTRL+SHIFT+ENTER. Remember to press this key combination whenever you enter or edit an array formula.
Other users may not understand your formulas. Array formulas are relatively undocumented, so if other people need to modify your workbooks, you should either avoid array formulas or make sure those users understand how to change them.
Depending on the processing speed and memory of your computer, large array formulas can slow down calculations.
Array Formula Heresy.
if you search in Column (A) use :
=INDIRECT("A" & SUMPRODUCT(MAX((A:A<>"")*(ROW(A:A)))))
if your range is A1:A10 you can use:
=INDIRECT("A" & SUMPRODUCT(MAX(($A$1:$A10<>"")*(ROW($A$1:$A10)))))
in this formula :
SUMPRODUCT(MAX(($A$1:$A10<>"")*(ROW($A$1:$A10))))
returns last non blank row number ,and indirect() returns cell value.
=INDEX(A:A, COUNTA(A:A), 1) taken from here
=MATCH("*";A1:A10;-1) for textual data
=MATCH(0;A1:A10;-1) for numerical data
Ive tried all the non-volatile versions but Not one version given above has worked.. excel 2003/2007update. Surely this can be done in excel 2003. Not as an array nor standard formula.
I either get just a blank, 0 or #value error.
So I resort to the volatile methods .. This worked..
=LOOKUP(2,1/(T4:T369<>""),T4:T369)
#Julian Kroné .. Using ";" instead of "," does NOT work! I think you are using Libre Office not MS excel?
LOOKUP is so annoyingly volitile I use it as a last resort only
For Microsoft office 2013
"Last but one" of a non empty row:
=OFFSET(Sheet5!$C$1,COUNTA(Sheet5!$C:$C)-2,0)
"Last" non empty row:
=OFFSET(Sheet5!$C$1,COUNTA(Sheet5!$C:$C)-1,0)
Place this code in a VBA module. Save. Under functions, User defined look for This function.
Function LastNonBlankCell(Range As Excel.Range) As Variant
Application.Volatile
LastNonBlankCell = Range.End(xlDown).Value
End Function
for textual data:
EQUIV("";A1:A10;-1)
for numerical data:
EQUIV(0;A1:A10;-1)
This give you the relative index of the last non empty cell in the range selected (here A1:A10).
If you want to get the value, access it via INDIRECT after building -textually- the absolute cell reference, eg:
INDIRECT("A" & (nb_line_where_your_data_start + EQUIV(...) - 1))
I had the same problem too. This formula also works equally well:-
=INDIRECT(CONCATENATE("$G$",(14+(COUNTA($G$14:$G$65535)-1))))
14 being the row number of the first row in the rows you want to count.
Chronic Clawtooth
I used HLOOKUP
A1 has a date;
A2:A8 has forecasts captured at different times, I want the latest
=Hlookup(a1,a1:a8,count(a2:a8)+1)
This uses a standard hlookup formula with the lookup array defined by the number of entries.
If you know that there are not going to be empty cells in between, the fastest way is this.
=INDIRECT("O"&(COUNT(O:O,"<>""")))
It just counts the non-empty cells and refers to the appropriate cell.
It can be used for a specific range as well.
=INDIRECT("O"&(COUNT(O4:O34,"<>""")+3))
This returns the last non empty cell in the range O4:O34.
This formula worked with me for office 2010:
=LOOKUP(2;1/(A1:A100<>"");A1:A100)
A1: the first cell
A100: refer to the last cell in comparing
I think the response from W5ALIVE is closest to what I use to find the last row of data in a column. Assuming I am looking for the last row with data in Column A, though, I would use the following for the more generic lookup:
=MAX(IFERROR(MATCH("*",A:A,-1),0),IFERROR(MATCH(9.99999999999999E+307,A:A,1),0))
The first MATCH will find the last text cell and the second MATCH finds the last numeric cell. The IFERROR function returns zero if the first MATCH finds all numeric cells or if the second match finds all text cells.
Basically this is a slight variation of W5ALIVE's mixed text and number solution.
In testing the timing, this was significantly quicker than the equivalent LOOKUP variations.
To return the actual value of that last cell, I prefer to use indirect cell referencing like this:
=INDIRECT("A"&MAX(IFERROR(MATCH("*",A:A,-1),0),IFERROR(MATCH(9.99999999999999E+307,A:A,1),0)))
The method offered by sancho.s is perhaps a cleaner option, but I would modify the portion that finds the row number to this:
=INDEX(MAX((A:A<>"")*(ROW(A:A))),1)
the only difference being that the ",1" returns the first value while the ",0" returns the entire array of values (all but one of which are not needed). I still tend to prefer addressing the cell to the index function there, in other words, returning the cell value with:
=INDIRECT("A"&INDEX(MAX((A:A<>"")*(ROW(A:A))),1))
Great thread!
If you are not afraid to use arrays, then the following is a very simple formula to solve the problem:
=SUM(IF(A:A<>"",1,0))
You must press CTRL + SHIFT + ENTER because this is an array formula.
INDEX returns a value by index position in an array and ROWS then is used to specify the last position of the array.
=LET(array,A1:A10,INDEX(array,ROWS(array)))
Also works for multiple columns when setting the parameter [column_num] of INDEX to 0:
=LET(array,A1:C10,INDEX(array,ROWS(array),0))
A simple one which works for me:
=F7-INDEX(A:A,COUNT(A:A))
Okay, so I had the same issue as the asker, and tried both top answers. But only getting formula errors. Turned out that I needed to exchange the "," to ";" for the formulas to work. I am using XL 2007.
Example:
=LOOKUP(2;1/(A:A<>"");A:A)
or
=INDEX(A:A;MAX((A:A<>"")*(ROW(A:A))))
For version tracking (adding the letter v to the beginning of the number), I found this one to work well in Xcelsius (SAP Dashboards)
="v"&MAX(A2:A500)