I gave this a search but could only find answers for when the row number is known.
I have a N by 4 matrix and am trying to remove rows based on the specific values in the second column and then the first column.
Initially I approached this the wrong way by filtering the matrix based on the values I didn't want, and then taking it away from the initial matrix, which obviously was not the result I wanted. (see below)
days = [669 680 298 299];
ind = ismember(B(:,2),days);
D = B(ind,:);
C=[B;-D];
I'm assuming there is a very similar way of removing rows, instead of filtering for them?
If you could help me in any way it would be much appreciated!
Edit:
Input:
1002 101 04 92
1002 12 12 298
1002 298 12 589
1002 680 12 589
Output:
1002 101 04 92
1002 12 12 298
I can't use row numbers as the matrix has millions of rows
To remove rows based on second column:
B(ismember(B(:,2),days),:) = [];
or equivalently
B = B(~ismember(B(:,2),days),:);
Or, in your code, change third and fourth lines into
C = B(~ind,:);
Related
Table copied as Text
Column1 Column2 Column3 Column4 Column5 Column6
A AA AAA 100 95 92
A AA AAA 85 83 81
A AA BBB 200 199 160
A BB AAA 65 55 49
B AA AAA 89 88 83
B AA BBB 150 149 145
B BB AAA 140 135
B BB BBB 190 185
B AA AAA 510
AA
AAA BBB
A 173 160
B 593 145
and some more explanation
Basically i want the sum of "Column 6" for the given criteria but the data in Column 6 can only be entered after some delay w.r.t. Column 1, Column 2, Column 3 & Column 4.
Till Column 6 data is entered, i want excel to use the number available in Column 5 which is also entered after some delay w.r.t. Column 1, Column 2, Column 3 & Column 4 but before Column 6.
And till Column 5 data is entered, i want excel to use the number available in Column 4.
Now I am familiar with two SUM/IF arrangements as included below in post.
First one is array sum/if arrangement which is convenient to write but results in terribly long calculation time with 1.5 seconds for just one column and I have over 100 columns in one sheet and about 9 sheets.
Second one is using SUMIFS which requires extensive time to write but relatively better calculation time of 0.5 seconds for column but is still quite high.
Now I need to do away with the array arrangement but doing so will take quite some time and I want to know if there is any better/other arrangement.
Just let me know other arrangement which can get the required result and I will check the arrangement for calculation timing. If the other arrangement is also convenient to write than that is a plus.
This is my table:
And I want to add the right most columns which are not empty i.e. have a number in it, but with the criteria for the first three columns in cell D15.
I only found option to add image. Please let me know how to upload excel file.
enter image description here
Can somebody please suggest an alternate to this array formula so it can calculate way faster
{=SUM(
IF(
($B$2:$B$10=$C15)*
($C$2:$C$10=$C$13)*
($D$2:$D$10=D$14)>0,
IF(
$G$2:$G$10<>"",
$G$2:$G$10,
IF(
$F$2:$F$10<>"",
$F$2:$F$10,
$E$2:$E$10))))}
I have tried below which reduces the calculation time to 1/3 but it is too much typing for the large data I am dealing with
=SUMIFS(
$G$2:$G$10,
$B$2:$B$10,$C15,
$C$2:$C$10,$C$13,
$D$2:$D$10,H$14,
$G$2:$G$10,"<>"&"")
+SUMIFS(
$F$2:$F$10,
$B$2:$B$10,$C15,
$C$2:$C$10,$C$13,
$D$2:$D$10,H$14,
$G$2:$G$10,"="&"",
$F$2:$F$10,"<>"&"")
+SUMIFS(
$E$2:$E$10,
$B$2:$B$10,$C15,
$C$2:$C$10,$C$13,
$D$2:$D$10,H$14,
$G$2:$G$10,"="&"",
$F$2:$F$10,"="&"")
If you're OK with using a helper column (which you should be), you can use this formula in a helper cell and drag down. (In my example at bottom, this formula is in cell H2 and drag down.)
= INDEX(E2:G2,MATCH(-1E+300,E2:G2,-1))
This gets all of the data in either column 4 5 or 6 all into one column.
Then you can use a simpler SUMIFS formula in cell D15:
= SUMIFS($H$2:$H$10, // Sum range (helper column)
$B$2:$B$10,$C15, // Criteria 1 (A or B)
$C$2:$C$10,$C$13, // Criteria 2 (AA or BB)
$D$2:$D$10,D$14) // Criteria 3 (AAA or BBB)
See below, working example:
DISCLAIMER
This answer will simplify your formulas, but I'm not sure if this will help with the performance problems you are experiencing. SUMIFS in itself I don't see being likely the cause of long calculation times. Probably you are experiencing long calculation times because other parts of your spreadsheet are using inefficient formulas and/or formulas involving volatile cells, but that is just a guess because I have no idea what the rest of your spreadsheet looks like.
I am new to array formulae and have noticed that while SUBTOTAL includes many functions, it does not feature COUNTIF (only COUNT and COUNTA).
I'm trying to figure out how I can integrate a COUNTIF-like feature to my array formula.
I have a matrix, a small subset of which looks like:
A B C D E
48 53 46 64 66
48 66 89
40 38 42 49 44
37 33 35 39 41
Thanks to the help of #Tom Shape in this post, I (he) was able to average the sum of each row in the matrix provided it had complete data (so rows 2 and 4 in the example above would not be included).
Now I would like to count the number of rows with complete data (so rows 2 and 4 would be ignored) which include at least one value above a given threshold (say 45).
In the current example, the result would be 2, since row 1 has 5/5 values > 45, and row 3 has 1 value > 45. Row 5 has values < 45 and rows 2 and 3 have partially or fully missing data, respectively.
I have recently discovered the SUMPRODUCT function and think that perhaps SUMPRODUCT(--(A1:E1 >= 45 could be useful but I'm not sure how to integrate it within Tom Sharpe's elegant code, e.g.,
=AVERAGE(IF(SUBTOTAL(2,OFFSET(A1,ROW(A1:A5)-ROW(A1),0,1,COLUMNS(A1:E1)))=COLUMNS(A1:E1),SUBTOTAL(9,OFFSET(A1,ROW(A1:A5)-ROW(A1),0,1,COLUMNS(A1:E1))),""))
Remember, I am no longer looking for the average: I want to filter rows for whether they have full data, and if they do, I want to count rows with at least 1 entry > 45.
Try the following. Enter as array formula.
=COUNT(IF(SUBTOTAL(4,OFFSET(A1,ROW(A1:A5)-ROW(A1),0,1,COLUMNS(A1:E1)))>45,IF(SUBTOTAL(2,OFFSET(A1,ROW(A1:A5)-ROW(A1),0,1,COLUMNS(A1:E1)))=COLUMNS(A1:E1),SUBTOTAL(9,OFFSET(A1,ROW(A1:A5)-ROW(A1),0,1,COLUMNS(A1:E1))))))
Data
I am reading data from a file and I am trying to do some manipulation on the vector containing the data basically i want to check if the values come from consecutive lines and if so i want to average each two and put the value in a output vector
part of the data and lines
lines=[153 152 153 154 233 233 234 235 280 279 280 281];
Sail=[ 3 4 3 1.5 3 3 1 2 2.5 5 2.5 2 ];
here is what i am doing
Sail=S(lines);
Y=diff(lines)==1;
for ii=1:length(Y)
if Y(ii)
output(ceil(ii/2))=(Sail(ii)+Sail(ii+1))/2;
end
end
is this correct also is there a way to do that without a for loop
Thanks
My suggestion:
y = find(diff(lines)==1);
output = mean([Sail(y);Sail(y+1)]);
This assumes that when you have, say [233 234 235], you want one value averaging the values from lines [233 234] and one value averaging those from [234 245]. If you wanted to do something more complex when longer sets of consecutive lines exist in your data, then the problem becomes more complex.
Incidentally it's a bad idea to do something like (ceil(ii/2)) - you can't guarantee a unique index for each matching value of ii. If you did want an output the same size as Sail (will have zeros in non-matching areas) then you can do something like this:
output2 = zeros(size(Sail));
output2(y)=output;
I am having difficulties with making an array formula work the way I want it to work.
Out of a column of dates which is not sorted, I want it to extract values into a new column. The formula below identifies the required cells of a given month and year, but they appear in their original row rather than on top of the output range. Moreover, I want all ""/FALSE cells to be excluded from the output array.
=IF((MONTH($I$15:$I$1346)=1)*(YEAR($I$15:$I$1346)=2008),$I$15:$I$1346,"")
In fact, the $I$15:$I$1346 should be dynamic and go to the last filled range (I could make a named range for that)
Part two is to expand on that formula so that it calculates the data that is an two column offset of the data described above.
Is the above possible to build into one cell probably with a combination of IF, INDEX, SMALL and maybe others?
I'm not looking for a filter solution. Hope the above is clear enough and that you can help!
Here's a shortened sample layout:
A B C
1 Date Series_A Series_B
2 03/01/2011 45 20
3 04/01/2011 73 30
4 06/01/2011 95 40
5 08/01/2011 72 50
6 06/02/2011 5 13
7 09/02/2011 12 #N/A
8 05/02/2011 23 65
9 07/03/2011 12 65
Then I want three input cells for the year and and the month and series name (index/match, as there are many more columns with data). If it would be 2011, Feb and Series_A, I want it to calculate the average for that month. In this case it would be (5+12+23)/3. If it would be Feb-2011 and Series_B instead, which has an error, it should show (13+65)/2 rather than an error.
Aside from that I want a separate which will output an array with the data instead without 'holes' in between and with the right 'length'. Example for Feb-2011 in Column C:
A B C D
1 Date Series_A Desired Output Output based on f above
2 03/01/2011 45 5
3 04/01/2011 73 12
4 06/01/2011 95 23
5 08/01/2011 72
6 06/02/2011 5 5
7 09/02/2011 12 12
8 05/02/2011 23 23
9 07/03/2011 12
If I then run a =ISBLANK(C5) it should be true, rather than =""=C5
Hope the edit clarifies
I reached out to various platsforms to get an answer, and here you have one which is ok. Still doesn't fully answer part 1, but works nonetheless.
http://www.excelforum.com/excel-formulas-and-functions/905356-exclude-blank-false-cells-in-in-excel-array-if-formula-output.html
I have an array comprising n rows and 4 colums. Each of the four entries on the row is an integer, i.e.,
X = [
111 112 432 2
6 9 115 111
112 432 111 2
];
Each row represents the vertices of a tetrahedron. These vertices have no directionality thus, in the case above, the tetrahedra represented by X(1,:) and X(3,:) are equivalent.
I wish to remove duplicate tetrahedra from X, but can't quite figure how to incorporate the order independence into my code.
I tried the UNIQUE() function but this returns a (nx1) array of unique integers, i.e.,
Y = UNIQUE(X);
Y = [
2
6
9
111
112
115
432
]
Anyone have any suggestions for a reasonably efficient way to complete this task?
Thanks,
S :-)
First, sort the rows of your matrix to arrive at a "canonical" representation for the tetrahedra:
X = sort(X, 2);
Then, use unique with the optional 'rows' argument to find unique rows:
Y = unique(X, 'rows');
unique() will work on rows, but rows 1 and 3 are a different order. So we could sort them prior to using unique.
Y=unique(sort(X,2),'rows')
Y =
2 111 112 432
6 9 111 115
If you want to retain the original ordering then unique will return the indices
[Y,yi]=unique(sort(X,2),'rows');
>> X(yi,:)
ans =
112 432 111 2
6 9 115 111
To quote from the documentation:
b = unique(A, 'rows') returns the
unique rows of A.
Is that what you want ?
you should sort the rows first, then use unique(A,'rows') as HPM suggests