比较两个字符串数组

Question

I have two lists of strings as a column in a table ( PM25_spr{i}.MonitorID and O3_spr{i}.MonitorID ). The lists are of different lengths. I want to compare the first 11 characters of each entry and pull out the index for each list where they are the same.

Example

List 1:
    '01-003-0010-44201'
    '01-027-0001-44201'
    '01-051-0001-44201'
    '01-073-0023-44201'
    '01-073-1003-44201'
    '01-073-1005-44201'
    '01-073-1009-44201'
    '01-073-1010-44201'
    '01-073-2006-44201'
    '01-073-5002-44201'
    '01-073-5003-44201'
    '01-073-6002-44201'

List 2:
    '01-073-0023-88101'
    '01-073-2003-88101'
    '04-013-0019-88101'
    '04-013-9992-88101'
    '04-013-9997-88101'
    '05-119-0007-88101'
    '05-119-1008-88101'
    '06-019-0008-88101'
    '06-029-0014-88101'
    '06-037-0002-88101'
    '06-037-1103-88101'
    '06-037-4002-88101'
    '06-059-0001-88101'
    '06-065-8001-88101'
    '06-067-0010-88101'
    '06-073-0003-88101'
    '06-073-1002-88101'
    '06-073-1007-88101'
    '08-001-0006-88101'
    '08-031-0002-88101'

I tried intersect , which isn't the right approach for what I want to do. I'm not sure how to use ismember given that I only want to look at the first 11 characters.

I tried strncmp , but Inputs must be the same size or either one can be a scalar.

chars2compare = length('18-097-0083'); 
strncmp(O3_spr{i}.MonitorID, PM25_spr{i}.MonitorID,chars2compare)

Answer 1

PM25_spr_MID = cell(length(years),1); % Preallocate cell array
for n = 1:length(PM25_spr{i}.MonitorID) 
    s = char(PM25_spr{i}.MonitorID(n)); % Convert string to char
    PM25_spr_MID{i}(n) = cellstr(s(1:11)); % Pull out 1-11 characters and convert to cell
end

O3_spr_MID = cell(length(years),1); % Preallocate cell array
for n = 1:length(O3_spr{i}.MonitorID)
    s = char(O3_spr{i}.MonitorID(n));
    O3_spr_MID{i}(n) = cellstr(s(1:11));
end

[C, ia, ib] = intersect(O3_spr_MID{i}, PM25_spr_MID{i}) 
PerCap_spr_O3{i} = O3_spr{i}(ia,:);
PerCap_spr_PM25{i} = PM25_spr{i}(ib,:);

Answer 2

Assuming list1 and list2 to be the two input cell arrays, you can use few approaches.

I. Operate on cell arrays

With intersect -

%// Clip off after first 11 characters in each cell of the input cell arrays
list1_f11 = arrayfun(@(n) list1{n}(1:11),1:numel(list1),'uni',0)
list2_f11 = arrayfun(@(n) list2{n}(1:11),1:numel(list2),'uni',0)

%// Use intersect to find common indices in the input cell arrays
[~,idx_list1,idx_list2] = intersect(list1_f11,list2_f11)

With ismember -

%// Clip off after first 11 characters in each cell of the input cell arrays
list1_f11 = arrayfun(@(n) list1{n}(1:11),1:numel(list1),'uni',0)
list2_f11 = arrayfun(@(n) list2{n}(1:11),1:numel(list2),'uni',0)

%// Use ismember to find common indices in the input cell arrays
[LocA,LocB] = ismember(list1_f11,list2_f11);
idx_list1 = find(LocA)
idx_list2 = LocB(LocA)

---

II. Operate on char arrays

We can use char dierctly on the input cell arrays to get 2D char arrays as working with them could be faster than working with cells .

With intersect + 'rows' -

%// Convert to char arrays
list1c = char(list1)
list2c = char(list2)

%// Clip char arrays after first 11 columns
list1c_f11 = list1c(:,1:11)
list2c_f11 = list2c(:,1:11)

%// Use intersect with 'rows' option
[~,idx_list1,idx_list2] = intersect(list1c_f11,list2c_f11,'rows')

---

III. Operate on numeric arrays

We can convert the char arrays further to numeric arrays with just one column as that could lead to faster solutions.

%// Convert to char arrays
list1c = char(list1)
list2c = char(list2)

%// Clip char arrays after first 11 columns
list1c_f11 = list1c(:,1:11)
list2c_f11 = list2c(:,1:11)

%// Remove char columns of hyphens (3 and 7 for the given input)
list1c_f11(:,[3 7])=[];
list2c_f11(:,[3 7])=[];

%// Convert char arrays to numeric arrays
ncols = size(list1c_f11,2);
list1c_f11num = (list1c_f11 - '0')*(10.^(ncols-1:-1:0))'
list2c_f11num = (list2c_f11 - '0')*(10.^(ncols-1:-1:0))'

This point onwards you have three more approaches to work with that are listed next.

With ismember ( would be memory efficient, but maybe not fast across all datasizes) -

[LocA,LocB] = ismember(list1c_f11num,list2c_f11num);
idx_list1 = find(LocA)
idx_list2 = LocB(LocA)

With intersect (could be slow) -

[~,idx_list1,idx_list2] = intersect(list1c_f11num,list2c_f11num)

With bsxfun ( would be memory inefficient, but maybe fast for small to decent sized inputs) -

[idx_list1,idx_list2] = find(bsxfun(@eq,list1c_f11num,list2c_f11num'))