How to iteratively apply a function to groups of data in Excel - excel

Question

So I'm quite new to using excel and it's possible I missed over a question like this that was already answered, but what I am trying to do is measure the distances between groups of points in a large excel sheet.
I've tried the standard way of writing a function and dragging it down but the way the function changes when I move it down is not what I need and I can't seem to figure how how I might fix this so I think a macro might be the solution.
The data is groups of 8 measurements with every group having one Blue, Red, Yellow, Csome, Overlap, Green, Cyan, and Full object. The XYZ coordinate of each object is listed in the first three columns and using a simple distance formula I am trying to measure the distance between each object and all the other objects in it's same group. Eg. Blue to Red, Yellow, Csome... etc. Then again for Blue2 to Red2, Yellow2, Csome2... and Blue3 to Red3, Yellow3, Csome3... and so fourth.
However, when I put the following formula in where my desired output should be and try to drag it down, obviously I run into some problems. For the first Blue against it's whole group, I begin with =SQRT((A$2-A2)^2+(B$2-B2)^2+(B$2-B2)^2) and drag it down so it cycles through row 2 versus rows 3,4,5,6,7,8,and 9. The issue arises when I get to row 10 which contains Blue2. This should not be compared with Blue1 but rather begins a new group. How then can I get the cycle to restart at row 10 so that the new group beginning with Blue 2 is compared within that group, eg the XYZ in row 10 to XYZ in 11, 12, 13...
:
Position X Position Y Position Z Unit Category Collection Surpass Object
16.38 41.20 4436.18 um Surface Position Blue Selection
17.35 40.83 4436.17 um Surface Position Red Selection
15.93 40.62 4435.93 um Surface Position Yellow Selection
16.22 40.80 4436.03 um Surface Position Csome1
17.97 42.23 4435.46 um Surface Position Overlap
15.37 40.87 4436.03 um Surface Position Green Selection
15.44 40.04 4436.39 um Surface Position Cyan Selection
17.97 42.23 4435.46 um Surface Position Full 1
15.57 36.20 4435.75 um Surface Position Blue Selection 2
12.83 35.19 4435.33 um Surface Position Red Selection 2
15.76 37.25 4435.74 um Surface Position Yellow Selection 2
14.94 36.17 4435.65 um Surface Position Csome2
15.99 36.78 4436.34 um Surface Position Overlap 2
14.23 36.26 4435.57 um Surface Position Green Selection 2
13.61 36.96 4436.09 um Surface Position Cyan Selection 2
14.64 36.42 4435.80 um Surface Position Full 2
42.83 36.04 4435.47 um Surface Position Blue Selection 3
42.34 36.56 4435.63 um Surface Position Red Selection 3
42.25 36.87 4435.75 um Surface Position Yellow Selection 3
42.48 36.50 4435.62 um Surface Position Csome3
41.77 36.02 4435.67 um Surface Position Overlap 3
42.64 36.89 4435.73 um Surface Position Green 3
42.05 36.90 4436.21 um Surface Position Cyan Selection 3
42.34 36.53 4435.71 um Surface Position Full 3
33.75 33.13 4436.32 um Surface Position Blue Selection 4
34.99 33.12 4436.38 um Surface Position Red Selection 4
35.40 31.97 4436.38 um Surface Position Yellow Selection 4
34.75 32.58 4436.36 um Surface Position Csome4
39.03 35.24 4436.04 um Surface Position Overlap Selection 4
35.49 32.19 4436.49 um Surface Position Green Selection 4
36.14 32.35 4436.42 um Surface Position Cyan Selection 4
36.16 33.20 4436.30 um Surface Position Full 4

Answer 1

Applying some VBA to loop through the matches to Full with a wildcard and reset the formula's starting row for each iteration should be the best way to go with this.
Sub Group_Formula()
Dim frmla As String, rw As Long, rws As Long, n As Long, cnt As Long
frmla = "=SQRT((A$×-A×)^2+(B$×-B×)^2+(B$×-B×)^2)"
With ActiveSheet '<-set to worksheet name properly!
For cnt = 1 To Application.CountIf(.Columns("G"), "Full*")
rw = Application.Max(2, rw)
rws = Application.Match("Full*", .Cells(rw, "G").Resize(Rows.Count - rw, 1), 0)
.Cells(rw, "H").Resize(rws, 1).Formula = Replace(frmla, Chr(215), rw)
rw = rw + rws
Next cnt
End With
End Sub
The results should resemble the following.
    

Answer 2

You need to handle the A$2, B$2 references with a function that will calculate the row. You can use reference functions like INDEX, INDIRECTor OFFSET to accomplish what you need. The last two are volatile, so keep in mind they recalculate every time a value changes in your sheet. INDEX is more efficient.
It isn't clear if your data includes the group number because the numbers at the end of the row are inconsistent. If they are consistent in your data you can use them. Supposing the group number is in column I, This INDEX formula will return the correct A$2 reference:
INDEX($A$2:$A$33,1+((I2-1)*8))
Or with the OFFSET function:
OFFSET($A$2,(I2-1)*8,0)
Otherwise, (without using the group number) you can use the ROWS function to produce a similar result:
INDEX($A$2:$A$33,1+QUOTIENT(ROWS(A$2:A2)-1,8)*8)
The 8 in all the formulas represent the number of items in a group.
And the full formula would be:
=SQRT((INDEX($A$2:$A$33,1+QUOTIENT(ROWS(A$2:A2)-1,8)*8)-A2)^2+(INDEX($B$2:$B$33,1+QUOTIENT(ROWS(B$2:B2)-1,8)*8)-B2)^2+(INDEX($C$2:$C$33,1+QUOTIENT(ROWS(C$2:C2)-1,8)*8)-C2)^2)

Answer 3

Well as a new excel user you've jumped straight into the wonderful and perplexing world of array formula's.
First you need to prep your data a bit. All those 1s, 2s, 3s and 4s need to be in a column of their own, all those in group 1 need a number, and everything needs to be named consistently ('Green' vs 'Green Selection'). We could make the formula even more complex to account for these idiosyncracies in your data, but cleaning it will make understanding it and the formulas easier.
Now we put the "Surpass Object" we want to calculate the distance from in Cell I1 (in this case 'Blue Selection') and the following formula into cell I2 being sure to hold ctrl+shift+enter when we enter it.
=SQRT((INDEX($A$2:$A$33,MATCH(1,($G$2:$G$33=I$1)*($H$2:$H$33=$H2),0))-$A2)^2+(INDEX($B$2:$B$33,MATCH(1,($G$2:$G$33=I$1)*($H$2:$H$33=$H2),0))-$B2)^2)
This will get us the distance between the x and y co-ordinates for each. You'll need to adjust the formula to add the z-coordinate to the calculation if you need it.
So what is the fancy index-match formula doing? In the first instance it grabs the cell in column a, that matches both the 'Surpass Object' in the column header (cell J$1) and the group of the current row (cell $H2)
Now you can just fill in the remaining headers and fill the formula right and down to get your matrices of differences