I am trying to recreate vlookup (the excel function) in matlab. Having trawled through numerous other questions, no solutions I have found do what I am looking for. A brief example is here. I have a set of objects each with a unique ID number. Database A contains a selection of IDs. Database B contains all of the IDs, plus additional attributes. I want to append the additional information per row from database B onto the end of database A.
So for example, if: A = [ 1; 2; 6; 7]
and B = [1 56.7; 2 45.1; 3 65.7; 4 76.1; 5 34.1; 6 34.1; 7 121.1]
I would like to change A to be A = [1 56.7; 2 45.1; 6 34.1; 7 121.1]
However, in my data I need to print the results from A in the 82nd column of B, and they need to be derived from the 8th column of A.
The following does not work for my purpose, because 'e' needs to change to be printed per row: http://www.mathworks.co.uk/matlabcentral/fileexchange/29233-vlookup-similar-to-ms-excel-function/content/vlookup.m
I also tried interp1 function in the following way based on previous results:
c=interp1(B(:,1),B(:,2),A(:,1))
But I need exact matches to my data, not interpolated values.
Any help much appreciated.

 채택된 답변

A Jenkins
A Jenkins 2014년 6월 9일

0 개 추천

Purely numeric and all index of A are guaranteed to be in B?
A = [ 1; 2; 6; 7];
B = [1 56.7; 2 45.1; 3 65.7; 4 76.1; 5 34.1; 6 34.1; 7 121.1];
A_col_id=1; % the column the ids are in matrix A
B_col_id=1; % the column the ids are in matrix B
A_col_value=2; % the column the values should go in matrix A
B_col_value=2; % the column the values are in matrix B
% real code
[~,locb]=ismember(A(:,A_col_id),B(:,B_col_id));
A(:,A_col_value)=B(locb,B_col_value)

댓글 수: 3

Dave
Dave 2014년 9월 24일
Following your answer, what if there are not 1 but 3 col_ids in A and B? The col_value are still in one column.
Dave, if you mean to say that the key is made of three columns, then you would use
ismember(A(:, A_col_keys), B(:, B_col_keys), 'rows')
Dave
Dave 2014년 9월 24일
Merci beaucoup. I'll check with the whole set but looks like suits perfectly.

댓글을 달려면 로그인하십시오.

추가 답변 (1개)

Max
Max 2015년 5월 2일

0 개 추천

For my own problem i found an answer without the usage of ismember but with a for loop also suitable for this problem:
A = [ 1; 2; 6; 7];
B = [1 56.7; 2 45.1; 3 65.7; 4 76.1; 5 34.1; 6 34.1; 7 121.1];
for i=1:length(A)
A(i,2)=B(B(:,1)==A(i,1),2);
end

댓글 수: 2

selina dong
selina dong 2015년 12월 8일
any idea how to include the criteria that when there is no corresponding match in B, fill with zero or NaN? At the moment it would give an error and would not run: 'Subscripted assignment dimension mismatch.'
thanks!
Guillaume
Guillaume 2015년 12월 8일
The answer is very simple, but please start your own question rather than hijacking somebody's else.

댓글을 달려면 로그인하십시오.

카테고리

질문:

2014년 6월 9일

댓글:

2015년 12월 8일

Community Treasure Hunt

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

Start Hunting!

Translated by