# Error in setdiff function

14 Ansichten (letzte 30 Tage)
Sagar Gupta am 7 Jul. 2021
Kommentiert: MEP am 25 Jan. 2022
Hi,
I have been tryin to use setdiff on two tables. There is a specific double column which contains NaN values and multiple rows which are same in both the tables. While using setdiff the rows that contains NaN in a specific column in both the tables comes as a difference between two tables, which should not happen. Both the rows are exactly same and the setdiff is considering NaN from same cells as different values. Is there a solution to this problem? Is there any other method to get the difference between the rows?
##### 0 Kommentare-2 ältere Kommentare anzeigen-2 ältere Kommentare ausblenden

Melden Sie sich an, um zu kommentieren.

### Akzeptierte Antwort

Bhavya Chopra am 8 Jul. 2021
I understand that you want to find the difference between two rows. NaN values are not considered equal, and the logical inequality test, (NaN ~= NaN), also returns true. The documentation for function setdiff specifies that it treats NaN values as distinct.
You might find the isequaln function to be useful to determine array equality, which treats NaN values as equal to each other, and returns a logical value.
As another work-around to Are Mjaavatten's answer, to obtain the difference between rows, you can also use the following approach:
a = [3 4 5 NaN NaN]; % Considering two example vectors
b = [3 NaN];
a_temp = a(~isnan(a)); % Removing NaN values using isnan() function
b_temp = b(~isnan(b));
setdiff(a_temp, b_temp) % Using setdiff to obtain difference
##### 1 Kommentar-1 ältere Kommentare anzeigen-1 ältere Kommentare ausblenden
MEP am 25 Jan. 2022
Hi, I have the same problem. My goal is to compare two tables and I want to use setdiff only that the NaN should be treated as the same and not as different. It is absurd that there isn't a dedicated option on the function to do this.

Melden Sie sich an, um zu kommentieren.

### Weitere Antworten (1)

Are Mjaavatten am 8 Jul. 2021
Bearbeitet: Are Mjaavatten am 8 Jul. 2021
One workaround is to replace all NaNs with some spceific value that is not present in your data, say -9999:
>> S1 = [1,2,3,NaN,5,6];S2 =[2,3,5,NaN];
>> setdiff(S1,S2)
ans =
1 6 NaN
>> S1(isnan(S1)) = -9999;S2(isnan(S2)) = -9999;
>> setdiff(S1,S2)
ans =
1 6
>> S1(S1==-9999) =NaN;S2(S2==-9999) = NaN; % Restore originals
##### 1 Kommentar-1 ältere Kommentare anzeigen-1 ältere Kommentare ausblenden
Are Mjaavatten am 8 Jul. 2021
Bearbeitet: Are Mjaavatten am 8 Jul. 2021
This function hopefully does what yout want:
function S = setdiffn(S1,S2)
dummy = rand;
while any(ismember(union(S1,S2),dummy))
dummy = rand; % Make sure dummy is not present in sets
end
S1(isnan(S1)) = dummy;S2(isnan(S2)) = dummy;
S = setdiff(S1,S2);
end

Melden Sie sich an, um zu kommentieren.

### Kategorien

Mehr zu Managing Data finden Sie in Help Center und File Exchange

R2021a

### Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by