SUMIFS with intermediate VLOOKUP in the criteria - excel

I have 3 tables, 1 of which I want to fill in columns with data based on the other 2. Tables are roughly structured as follows:
Table 1 (Semi-Static Data)
SubGroup Group
----------- -----------
subgroup(1) group(a)
subgroup(2) group(b)
subgroup(3) group(b)
subgroup(4) group(c)
etc.
Table 2 (Variable Data)
SubGroup DataValue
----------- -----------
subgroup(1) datavalue(i)
subgroup(2) datavalue(ii)
subgroup(3) datavalue(iii)
subgroup(4) datavalue(iv)
etc.
Table 3 (Results)
Group TotalValue
----------- -----------
group(a) totalvalue(m)
group(b) totalvalue(n)
group(c) totalvalue(o)
etc.
Where the TotalValue is the sum of all DataValue's for all subgroups that belong to that particular Group.
e.g. for group(b) ---> totalvalue(n) = datavalue(ii) + datavalue(iii)
I am looking to achieve this calculation without adding any additional columns to the Data tables nor using VBA.
Basically I need to perform a COUNTIFS where there is an additional VLOOKUP matching the subgroup criteria range to the group it belongs to, and then only summing for datavalue's that match the group being evaluated. I have tried using array formulas but I'm having issues making it work. Any assistance would be very appreciated. Thank you,
EDIT: Wanted to add some details surrounding my question. First all Google searches did not provide a suitable answer. All the links had solutions to a slightly different problem were the VLOOKUP term is not dependent on the SUMIFS criteria but rather another single static variable. Stack Overflow offered similar solutions. Please let me know if anymore details are required to make my post suitable for this forum. Thank you again.

You can use the SUMPRODUCT function to do it all at once. The first reference $B$2:$B$5 is for the Group names, the second reference $E$2:$E$5 is for the datavalues. The G2 reference is for the group names in the third table, you can enter this formula for the first reference and then drag and fill for the rest.
=SUMPRODUCT($E$2:$E$5 * (G2 = $B$2:$B$5))

Some cell references, and sample data, would be helpful but something like this might be what you want:
=SUMIF(C:C,"="&INDEX(A:A,MATCH(E5,B:B,0)),D:D)

WADR & IMHO, this is simply bad worksheet design. For lack of a single cross-reference column in Table2, any solution would have to be a VBA User Defined Formula or an overly complicated array formula (the latter of which I am not even sure is possible). The data tables are not normalized database tables you can INNER JOIN or GROUP BY ... HAVING.
The formula you are trying to achieve is akin to,
=SUMPRODUCT(SUMIF(D:D, {"subgroup(2)","subgroup(3)"}, E:E))
That only works with hard-coded values as arrayed constants (e.g. {"subgroup(2)","subgroup(3)"}). I know of no way to spit a dynamic list back into the formula using additional native Excel functions but VBA offers some possibilities.
HOWEVER,
The simple addition of one more column to Table2 with a very basic VLOOKUP reduces all of your problems to a SUMIF.
     
The formula in the new column D, row 2 is,
=VLOOKUP(E2, A:B, 2, FALSE)
The formula in I2 is,
=SUMIF(D:D, H2,F:F )
Fill each down as necessary. Sorry if that is not what you wanted to hear.

Thank you everyone that responded and reviewed this post. I have managed to resolve this using an array formula and some matrix algebra. Please note that I am not using VLOOKUP (this operator cannot be performed on arrays) nor SUMIFS as my title states.
My final formula looks like this:
{=SUM(IF([Table2.xlsx]Sheet1!SubGroup=TRANSPOSE(IF([Table1.xlsx]Sheet1!Group=G2,[Table1.xlsx]Sheet1!SubGroup,"")),[Table2.xlsx]Sheet1!DataValue))}
Very simply, I create an array variable that compares the Group being evaluated (e.g. cell G2) with the Groups column for Table 1 and outputs the corresponding matching SubGroups. This results in an array with as many rows as Table 1 had (N) and 1 column: Nx1. I then transpose that array (1xN) and compare it to the SubGroups column (Mx1, M being the number of rows in Table 2) and output the DataValues column for the rows that have a corresponding SubGroup (MxN). Then I perform a sum of the whole array to return a single value.
Notice that as I didn't include a value_if_false output return on either IF operators, it will just populate with FALSE in the arrays were the conditions are not met. This does not matter though for the final result. In the first IF, FALSE will not match the SubGroups so will be ignored. For the second all values FALSE passed to SUM will be calculated as 0. The more complicated question is that it grows the amount of memory required to process as we are not filtering to just have the values we want.
For this application I decided against filtering the subarray as the trade-off in resource utilization was acceptable. If the data sets were any bigger though, I would definitely try doing it. Another concern was that I did not understand fully the filtering logic that I was using based on http://exceltactics.com/make-filtered-list-sub-arrays-excel-using-small/ so decided to simplify. Will revisit this concept latter as I think it will work. I might have completed this solution but was missing transposing the array to compare properly so abandoned this route.

Related

Rank equally across multiple columns with different data types using a single formula. Is it possible?

See this data set below. It has 3 columns (PTQ, % Growth, $ Growth) that all need ranked individually then summed up and then ranked again for a total power rank of each region. Is there any way I can do this with a single formula? I do this a lot and it would be nice not to have to rank everything individually each time. Basically I need to end up at the "Power Rank" column.
To clarify, I do not want to rank first on one column then another, they all need to be ranked equally together.
Thought I'd add a solution for non-365 users:
Edit: slight amendment so that the returned ranking mimics that of the RANK function, with thanks to #Scott Craner
=MMULT(N(MMULT(CHOOSE({1,2,3},RANK(A2:A10,A2:A10),RANK(B2:B10,B2:B10),RANK(C2:C10,C2:C10)),{1;1;1})>TRANSPOSE(MMULT(CHOOSE({1,2,3},RANK(A2:A10,A2:A10),RANK(B2:B10,B2:B10),RANK(C2:C10,C2:C10)),{1;1;1}))),ROW(A2:A10)^0)+1
which may require committing with CTRL+SHIFT+ENTER, depending on the version of Excel being used.
The static
{1;1;1}
relates to the fact that 3 columns are being queried, and could of course be replaced with a dynamic alternative.
We can use Office 365 LET and SORT. RANK does not allow the use of arrays, so we need to improvise with MATCH and SORT.
=LET(
ptq,A2:A10,
grwt,B2:B10,
grwthd,C2:C10,
ptqrnk,RANK(ptq,ptq),
gwtprnk,RANK(grwt,grwt),
gwtdrnk,RANK(grwthd,grwthd),
sm,ptqrnk+gwtprnk+gwtdrnk,
MATCH(sm,SORT(sm),0))

Excel multilevel array formula with partial string matches to sum resultant cells

I've been trying to sort this for over a day now without much luck. I have successfully used SUMIFS, INDEX, MATCH, COUNTIF, "--" etc array functions previously and am not a novice, but also not an expert on these. I can't seem to weave these together correctly, and likely on an altogether incorrect path.
Basically, I am trying to aggregate data from multiple spreadsheets, requiring a mapping of various items (rows) into a canonical form for summing.
The image here shows a representative, but simplified version of my quest. Each "region" on this example spreadsheet (Final..., Mapping, DataSet1, DataSet2) is actually in different spreadsheets, and there are several sheets with 50-150 rows in each xlsx.
Note that the names in Column B are quite arbitrary (meaning not all P1's have an 'x' pattern, like shown here as x1, x2, etc. Do not rely on any pattern in the names, except the x, y , z in the Mapping table are substrings (case insensitive, trailing match) of the names in Column B in the DataSets.
And in the image, the Final Result Table (summed manually) is what I want to compute via(an array) formula: A single formula would be ideal (given I have many spreadsheets from which the monthly data is being pulled from, so I can't readily modify but can create an interim spreadsheet if required, so open to helper columns or helper rows).
Here's the process - For each name (B3-B5) in the Final Result Table, I want to sum the name from it's components as follows:
Lookup all the matches in the Mapping Table (so for P1, the formula =IF($C$10:$C$15=$B3, $B$10:$B$15,"") gives {"x1";"";"";"x2";"";"x3"}.
I then want to search each of x1, x2, and x3 in B19:B26 to get rows 21, 22, 24, 25, 26 in DataSet1 and B31:B35 to get row 32 in DataSet2, to then add up the Jan totals into C3. (Effectively,
C3=C21+C22+C24+C25+C26+C32). Same for P2 and P3, and thru Feb, Mar, ...
I am stuck on how to remove blank or 0 or Div0 or such "error rows" from the interim result in 2, and also need to use 2 arrays of different sizes (3 valid rows in example 2 above, ignoring blanks) to search many rows in DataSets. I tried SEARCH("*"&IF($C$10:$C$15=$B3, $B$10:$B$15,""), $B$19:$B$26) but get unexpected results. I have tried to replace text in the interim result {"x1";"";"";"x2";"";"x3"} with TRUE/FALSE, and 1/0, etc. to help with INDEX or MATCH, but am stymied by errors in downstream ("surrounding") formulas.
Thanks in advance.
Here is a solution without resorting to nasty (imo) CSE formulas.
= SUMPRODUCT($C$19:$F$26*(COUNTIFS($B$10:$B$15, RIGHT($B$19:$B$26,2),$C$10:$C$15,$B3)>0)*($C$18:$F$18=C$2))
+
SUMPRODUCT($C$31:$F$35*(COUNTIFS($B$10:$B$15, RIGHT($B$31:$B$35,2),$C$10:$C$15,$B3)>0)*($C$30:$F$30=C$2))
There is one SUMPRODUCT for each data set. If possible, it would be better to put all your data sets into a single table with a column identify which data set it is a part of.
The way it works is to takes each values in your data set and multiplies it by whether the 2 right most character appear in your mapping table for that P code, multiplied by whether the value is in the correct month. So it returns 0 if either of those conditions are false. Then returns the sum.
UPDATE IN RESPONSE TO OP COMMENTS
If, the X,Y, Z codes are not always 2 digits but the first part is ALWAYS 8 digits, you can easily amend the:
RIGHT($B$19:$B$26,2)
to be:
RIGHT($B$19:$B$26,LEN($B$19:$B$26)-8)
Making the formula for the first data set:
=SUMPRODUCT($C$19:$F$26*(COUNTIFS($B$10:$B$15, RIGHT($B$19:$B$26,LEN($B$19:$B$26)-8),$C$10:$C$15,$B3)>0)*($C$18:$F$18=C$2))
And you can amend for other data sets and simply add them together.
Nice challenge! Are you willing to drop all your tables (DataSet1, DataSet2...) into one spreadsheet, so that we can refer just one single range for each month?
Here's one solution (hopefully a good starting point) - array formula (Ctrl+Shift+Enter):
=SUMPRODUCT(IFERROR(IF(TRANSPOSE(IF($B3=$C$10:$C$15,$B$10:$B$15,""))=RIGHT($B$18:$B$36,2),C$18:C$36,0),0))

Ranking with subsets

I'm trying to rank values and have managed to work out how to sort ties. My data looks at the total number of entries, ranks based on that and if there is a tie it looks to the next column of values to sort them out. However, I have two classes (East and West I've called them) of data within my dataset and want to rank them both separately (but stick to the rules above). So, if I had seven entries, 3 of them West and 4 of the East, I want West to have ranking 1,2,3 based on all the values that lie in that subset and East would have ranking 1,2,3,4. Can you explain what your formula is doing so I can understand how to apply your answer better in the future.
Effectively I'm asking what formula needs to go in achieve my result.
Cheers
Paul
There are a few related ways to do this, most involving SUMPRODUCT. If you don't like the solution below and would like to research other ways/explanations, try searching for "rankif".
The function looks up the Class and Value columns and, for every value in those columns, returns a TRUE or 1 if the current Class is a match AND if its Value is larger than the current Value, False or 0 if otherwise. The SUM adds up all these 1s, and the 1+ is for decoration. Remember to enter as an array formula using Ctrl+Shift+Enter before dragging down.
I used the array formula and SUM above to explain, but the following also works and might even be faster since it's not an array formula. It's the same idea, except we hijack SUMPRODUCT's ability to spit out a single value from an array.
=1+SUMPRODUCT(($A$2:$A$8=A2)*($B$2:$B$8>B2))
EDIT
To extend the rank-if, you could add more subsets to rank by multiplying more conditions:
You can also easily add tiebreakers by adding another SUMPRODUCT to treat the ties as an additional subset:
The first SUMPRODUCT is the 'base rank', while the second SUMPRODUCT is tiebreaker #1.

nested excel functions with conditional logic

Just getting started in Excel and I was working with a database extract where I need to count values only if items in another column are unique.
So- below is my starting point:
=SUMPRODUCT(COUNTIF(C3:C94735,{"Sharable Content Object Reference Model 1.2","Authored SCORM/AICC content","Authored External Web Content"}))
what i'd like to figure out is the syntax to do something like this-
=sumproduct (Countif range1 criteria..., where range2 criteria="is unique value")
Am I getting this right? The syntax is a bit confusing, and I'm not sure I've chosen the right functions for the task.
I just had to solve this same problem a week ago.
This method works even when you can't always sort on the grouping column (J in your case). If you can keep the data sorted, #MikeD 's solution will scale better.
Firstly, do you know the FREQUENCY trick for counting unique numbers? FREQUENCY is designed to create histograms. It takes two arrays, 'data' and 'bins'. It sorts 'bins', then creates an output array that's one longer than 'bins'. Then it takes each value in 'data' and determines which bin it belongs in, incrementing the output array accordingly. It returns the array. Here's the important part: If a value appears in 'bins' more than once, any 'data' value meant for that bin goes in the first occurrence. The trick is to use the same array for both 'data' and 'bins'. Think it through, and you'll see that there's one non-zero value in the output for each unique number in the input. Note that it only counts numbers.
In short, I use this:
=SUM(SIGN(FREQUENCY(<array>,<array>)))
to count unique numeric values in <array>
From this, we just need to construct arrays containing numbers where appropriate and text elsewhere.
In the example below, I'm counting unique days when the color is red and the fruit is citrus:
This is my conditional array, returning 1 or true for the rows I'm interested in:
($A$2:$A$10="red")*ISNUMBER(MATCH($B$2:$B$10,{"orange","grapefruit","lemon","lime"},0))
Note that this requires ctrl-shift-enter to be used as an array formula.
Since the value I'm grouping by for uniqueness is text (as is yours), I need to convert it to numeric. I use:
MATCH($C$2:$C$10,$C$2:$C$10,0)
Note that this also requires ctrl-shift-enter
So, this is the array of numeric values within which I'm looking for uniqueness:
IF(($A$2:$A$10="red")*ISNUMBER(MATCH($B$2:$B$10,{"orange","grapefruit","lemon","lime"},0)),MATCH($C$2:$C$10,$C$2:$C$10,0),"")
Now I plug that into my uniqueness counter:
=SUM(SIGN(FREQUENCY(<array>,<array>)))
to get:
=SUM(SIGN(FREQUENCY(
IF(($A$2:$A$10="red")*ISNUMBER(MATCH($B$2:$B$10,{"orange","grapefruit","lemon","lime"},0)),MATCH($C$2:$C$10,$C$2:$C$10,0),""),
IF(($A$2:$A$10="red")*ISNUMBER(MATCH($B$2:$B$10,{"orange","grapefruit","lemon","lime"},0)),MATCH($C$2:$C$10,$C$2:$C$10,0),"")
)))
Again, this must be entered as an array formula using ctrl-shift-enter. Replacing SUM with SUMPRODUCT will not cut it.
In your example, you'd use something like:
=SUM(SIGN(FREQUENCY(
IF(ISNUMBER(MATCH($C$3:$C$94735,{"Sharable Content Object Reference Model 1.2","Authored SCORM/AICC content","Authored External Web Content"},0)),MATCH($J$3:$J$94735,$J$3:$J$94735,0),""),
IF(ISNUMBER(MATCH($C$3:$C$94735,{"Sharable Content Object Reference Model 1.2","Authored SCORM/AICC content","Authored External Web Content"},0)),MATCH($J$3:$J$94735,$J$3:$J$94735,0),"")
)))
I'll note, though, that scaling might be a problem on data sets as large as yours. I tested it on larger data sets, and it was fairly fast on the order of 10k rows, but really slow on the order of 100k rows, such as yours. The internal arrays are plenty fast, but the FREQUENCY function slows down. I'm not sure, but I'd guess it's between O(n log n) and O(n^2) depending on how the sort is implemented.
Maybe this doesn't matter - none of this is volatile, so it'll just need to calculate once upon refreshing the data. If the column data is changing, though, this could be painful.
Asuming the source data is sorted by the key value [A], start with determining the occurence of the key column
B2: =IF(A2=A1;B1+1;1)
Next determine a group sum
C2: =SUMIF($A$2:$A$9;A2;$B$2:$B$9)
A key is unique if its group sum is exactly 1
D2: =(C2=1)
To count records which match a certain criterium AND are unique, include column D in a =IF(AND(D2, [yourcondition];1;0) and sum this column
Another option is to asume a key unique within a sorted list if it is unequal to both its predecessor and successor, so you could find the unique records like
E2: =AND(A2<>A1;A2<>A3)
G2: =IF(AND(E2;F2="this");1;0)
E and G can of course be combined into one single formula (not sure though if that helps ...)
G2(2): =IF(AND(AND(A2<>A1;A2<>A3);F2="this");1;0)
resolving unnecessarily nested AND's:
G2(3): =IF(AND(A2<>A1;A2<>A3;F2="this");1;0)
all formulas in row 2 should be copied down to the end of the list

Excel: How to parse/cast text as a formula?

Is it possible to parse/cast text (like "=A1+A2") as a formula in MS Excel? I want to build a formula from pieces of text - some of which will only be typed in later by a user.
If the INDIRECT() function did not only work for referencing cells, then I could have typed this =INDIRECT("=A1+A2").
I know you can a work around this problem by simply adding a lot more hidden columns to do sub calculations. But for the sake scalability and efficiency, I would rather do something like the above.
I found a similar questions here and here, yet they don't solve my problem.
The Real-world problem:
Read on for a better understanding as to why you would want to do the above
Scenario
Each item in the list consists of a string, which contains anywhere from 1 to 5 account names each. Each account name is followed by an account number in brackets. The length of the number determines the type of account. Part of the account number is a date, of which the date format depends on the type of account. Further more, each account type may have more that 1 account-number length associated with it, although each number-length[*] is only associated with 1 account type.
Objectives
Extract account-names and their respective account-numbers and account-types from a list.
Make an assumption as to the account-type from the account-number
Validate this assumption by inspecting the build of the number and elements in the name
Check the validity of the account-numbers depending on their type.
The tricky part (this is where my problem lies)
The account-types and their respective account-number-lengths are not known before hand, and are typed into a table by the user of the sheet, specifying a type of account and the number-lengths associated with this account-type. The user should type this into a list - not go and tinker around with delicate formulas
Done so far
Column A: Contains the raw data (each cell has up to 5 names and numbers)
Columns B..F: Each column extracts 1 name, remains empty if all are already extracted
Columns G..K: Each column extracts 1 number corresponding to its name in columns B..F, remains empty if all are already extracted
Columns L..P: Each column calculates the length of the corresponding number in columns
G..K
Now the user would type the following details into a table which assigns certain number-lengths an account type:
TYPE2, BUSINESS, (OR(length=13,length=6))
where length will later be replaced with the cell address which contains the calculated account number-length.
What I want to do now
Columns Q..U:
Should all indicate the account-type of the corresponding account-number in columns G..K. The idea is to build a nested if-elseIf-elseIf formula using the criteria typed in by the user as specified above. Example of one of the elseIF statements:
SUBSTITUTE(CONCATENATE("=IF(",criteria,",",type,",",errCode)),"length","O10"))
All of these elseIf statements will then be concatenated together to form a master formula which will then need to be parsed/cast as a formula to calculate the account-type
This proposal uses only 5 columns (1 for each account-number, containing the master formula) and a table specifying account-types and criteria, also keeping the user away from formulas. Editing 1 line of code (the criteria) will update all formulas. Efficient & Scalable.
Since the user should never tinker around with the formulas under the hood, a simple 1 column if-elseIf-elseIf is out of the question. The alternative to the above would be to make a separate column to test for each account-type for each account-number. Separating/Abstracting out each test to its own column results in much better readability, easier editing & much less debugging - Unless you like multi-screen-wide-formulas. Example: 5 account-numbers * 10 possible account types = 50 extra columns.
Each edit to any criteria needs to copied to 4 other non-adjacent columns and drag-filled down 10,000 rows (columns can not be adjacent since it is effectively a 5x5 array of columns). Not Efficient nor scalable. Unless I'm missing some elegant way of updating non-adjacent formulas in a single click
The rest of the validations error indications are trivial.
Sample data
Tshepo Trust (6901/2005) Marlene Mead (8602250646085)
Great Force Inv 67 Pty Ltd (200602258007)
Jane (870811) Livingstone (6901/2005) Janette Appel (8503250647056) James (900111)
I know all this would probably be much easier to achieve with clever usage of VBA, eliminating all the need to simulate abstraction, encapsulation, multi-dimensional arrays and functional programming on a spreadsheet. But until I can program in VBA, worksheet formulas will be my refuge.
[*]: account number-length could also be described as the amount of digits in the number or as indicated by this formula: LEN(accNumber)
In VBA you have access to Cell.Formula.
I usually used Range to peek a cell by address.
I'm not sure if this would answer your question(it's a very detailed question!), but if your user was entering the account numbers in a table (I'm calling it 'RefTable') , that was:
Length of account number | business type
----------------------------------------
6 | Accountant
8 | Advisor
Then you could just use a vlookup on the length of the account number, given you've already separated them out.
=vlookup(len(accNumber), Reftable, 2, false)
Make sure that you either use a dynamic range name, or specify plenty of space below in RefTable, so that when your users add types, they don't get lost.
Also, if you have two different accounts with the same length, this could get you into trouble.

Resources