How can I speed up this indexing code?
5 Ansichten (letzte 30 Tage)
Ältere Kommentare anzeigen
bhousden
am 28 Okt. 2021
Kommentiert: bhousden
am 29 Okt. 2021
I have a cell array in which each cell contains a string. E.g....
a={'AAAAAA';'BBBBBB';'CCCCCC';'AAAAAA';'DDDDDD'};
Each cell in array a is associated with a row in a numerical array that contains 10 columns. E.g....
b=[0,0,0,1,1,0,0,0,0,0;
0,1,0,0,0,0,0,0,0,0;
1,0,1,0,1,0,2,0,0,0;
3,0,0,0,0,0,0,0,0,1;
0,0,0,0,0,0,2,0,0,1];
Some strings in a are repeated such as 'AAAAAA' as shown above. What I need to do is find all repeated cells in a and sum the assocated columns from b into a single row. This should result in two new arrays (unia and bnew) which have equal numbers of rows but every string in unia is unique.
Easy enough to do with a loop such as:
unia=unique(a);
bnew=zeros(numel(unia),10);
for n=1:numel(unia)
pos=find(strcmp(a,unia{n}));
bnew(n,:)=sum(b(pos,:),1);
end
This works fine for small arrays but I have a case where a has 6 million cells and unia has 300,000 cells so I need something much faster. Any ideas?
Thanks!
0 Kommentare
Akzeptierte Antwort
Ive J
am 28 Okt. 2021
Avoid comparing strings within the loop and instead take advantage of the index vector from unique:
a = ["A", "B2", "A", "C", "AA", "B2", "B2"]; % use strings instead of cell array of characters, they're much more efficinet to work with
b = randi([0 2], numel(a), 3)
[anew, ~, idx] = unique(a);
bnew = arrayfun(@(x) sum(b(x == idx, :), 1), 1:numel(anew), 'uni', false);
bnew = vertcat(bnew{:})
anew
Also, you can use tall arrays when dealing with large arrays.
Weitere Antworten (0)
Siehe auch
Kategorien
Mehr zu Matrix Indexing finden Sie in Help Center und File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!