在单元格内检查Ismember

Question

我有一个单元格数（data），看起来像这样（在此处缩短）：

'45.203885' '-90.600123'    '119-8001'  733144  NaN
'45.203885' '-90.600123'    '119-8001'  733147  NaN
'45.203885' '-90.600123'    '119-8001'  733150  NaN
'45.203885' '-90.600123'    '119-8001'  733153  NaN
'45.203885' '-90.600123'    '119-8001'  733156  NaN
'45.203885' '-90.600123'    '119-8001'  733159  NaN

我想填写NaN的第5列，具体取决于第四列（使用datenum转换的日期）是否匹配B。

B（也是一个单元格）看起来像这样（示例也很缩短了很多）：

'45.203885' '-90.600123'    '119-8001'  733144  '3.3'
'45.203885' '-90.600123'    '119-8001'  733150  '9.5'
'45.203885' '-90.600123'    '119-8001'  733156  '6.8'

您可以看到，第四列日期在B中不会始终如一。我正在尝试将NaN'S添加到第5列中，其中B(:,3)和B(:, 4)与data(:,3)和data(:, 4)不匹配。

最终产品应该看起来像：

'45.203885' '-90.600123'    '119-8001'  733144  '3.3'
'45.203885' '-90.600123'    '119-8001'  733147  NaN
'45.203885' '-90.600123'    '119-8001'  733150  '9.5'
'45.203885' '-90.600123'    '119-8001'  733153  NaN
'45.203885' '-90.600123'    '119-8001'  733156  '6.8'
'45.203885' '-90.600123'    '119-8001'  733159  NaN

如果 data是矩阵，我只会执行以下操作：

data_ind = ismember(data(:,3:4),B(:,3:4),'rows');

，但我不知道如何使用单元格。某种形式的cellfun会做技巧吗？

Answer 1

sd = size(data,1); %// number of rows of data
sb = size(B,1); %// number of rows of B
[dd bb] = ndgrid(1:sd,1:sb); %// all combinations (row of data, row of B)
cond1 = strcmp(data(dd,3),B(bb,3)); %// test col 3 for all combinations
cond2 = [data{dd,4}].'==[B{bb,4}].'; %// test col 4 for all combinations
cond = reshape(cond1 & cond2, sd, sb); %// combine the two conditions
[ib, id] = max(cond); %// id contains the index of the first 1 (if any) ...
%// ... of each col in cond; and ib is a logical index of the row of that 1
id = id(ib); %// keep only id for which the maximum is 1
data(id,:) = B(ib,:); %// copy matching rows of B into data

示例data和B都包含与其他变量的任何行不匹配的行：

data = {
    '45.203885' '-90.600123'    '119-8001'  733144  NaN
    '45.203885' '-90.600123'    '119-8001'  733147  NaN
    '45.203885' '-90.600123'    '119-8001'  733150  NaN
    '45.203885' '-90.600123'    '119-8001'  733153  NaN
    '45.203885' '-90.600123'    '119-8001'  733156  NaN
    '45.203885' '-90.600123'    '119-8001'  733159  NaN};
B = {
    '45.203885'    '-90.600123'    '119-8001'    [733144]    '3.3'
    '45.203885'    '-90.600123'    '119-8001'    [733150]    '9.5'
    '45.203885'    '-90.600123'    '119-8001'    [733156]    '6.8'
    '45.203885'    '-90.600123'    '169-8001'    [833156]    '6.8'};

结果：

data = 
    '45.203885'    '-90.600123'    '119-8001'    [733144]    '3.3'
    '45.203885'    '-90.600123'    '119-8001'    [733147]    [NaN]
    '45.203885'    '-90.600123'    '119-8001'    [733150]    '9.5'
    '45.203885'    '-90.600123'    '119-8001'    [733153]    [NaN]
    '45.203885'    '-90.600123'    '119-8001'    [733156]    '6.8'
    '45.203885'    '-90.600123'    '119-8001'    [733159]    [NaN]

Answer 2

目前尚不清楚您对第三列做什么，因为所有条目都是相同的。另外，您的问题有些困惑，因为您想要data中的B还是B中的data。可能最快，最直接的方法是使用for循环：

data = {'45.203885' '-90.600123'    '119-8001'  733144  NaN
        '45.203885' '-90.600123'    '119-8001'  733147  NaN
        '45.203885' '-90.600123'    '119-8001'  733150  NaN
        '45.203885' '-90.600123'    '119-8001'  733153  NaN
        '45.203885' '-90.600123'    '119-8001'  733156  NaN
        '45.203885' '-90.600123'    '119-8001'  733159  NaN};
B = {'45.203885' '-90.600123'    '119-8001'  733144  '3.3'
     '45.203885' '-90.600123'    '119-8001'  733150  '9.5'
     '45.203885' '-90.600123'    '119-8001'  733156  '6.8'};
d3 = data(:,3);
d4 = [data{:,4}].';
for i = 1:size(B,1)
    data(strcmp(d3,B{i,3})&d4==B{i,4},5) = B(i,5);
end

不要害怕使用for循环。您也可以使用cellfun进行操作，但是它需要使用eval。

Answer 3

，由于您只有唯一的ID Datenum组合，因此您可以执行以下操作：

data_ind = ismember(strcat(data(:,3),num2str([data{:,4}]')),...
                    strcat(B(:,3),num2str([B{:,4}]')));

这所做的是将第三列和第4列加入到一个字符串中，例如

'45.203885' '-90.600123'    '119-8001'  733144  NaN
'45.203885' '-90.600123'    '119-8001'  733147  NaN

将成为

'119-8001733144'
'119-8001733147'

等等。然后，它将这些字符串从data矩阵与B矩阵进行比较，从而为您提供了索引矩阵。