To delete rows from a cell array specifying conditions

I was searching for smth like this, but I haven't found it. The task is: I have a cell array with 25 columns. I need to filter data according to some conditions - only with FLAGS 'O PGD', with SoT 'track' and without 'nan'. So, as a result I need to delete rows which contain this info and corresponding lat, long, so that only reliable data left.

I've written a part of the program to filter data:

A=raw(:,3); %PGD flags
B=raw(:,17); %track,search,memory modes
D=raw(:,14); %distance
C=find(strcmp(' O PGD ',raw(:,3))); %indeces
C1=D(C); %values of distance
E=find(strcmp('track',B)); %indeces
E1=D(E); %values of distance
But, in this case, only distance is filtered and I can't delete another rows(lat,long).
Also, I was trying to work with num data (I mean, to represent the same information, but in num format). I've deleted only rows with 'nan', but in the whole file:
TF1=isnan(num2(:,14)); %condition, build in function
num2(TF1,:)=[]; %delete rows which specify condition in file num2
So, the question will be: how can I delete rows in cell format specifying these conditions OR how to write a condition in num format to delete these rows?

Example of information:

'Flags' 'Dist[km]' 'SoT' 'Lat.' 'Long.'
' O PGD ' 'nan' 'search' 47.562701 2.169937
' O PGD ' 'nan' 'search' 47.562661 2.170009
' O PGD ' 'nan' 'track' 47.562622 2.170081
' O PGD ' 66.4 'track' 47.562584 2.170151
' O PGD ' 66.4 'track' 47.562544 2.170225
' O PGD ' 66.4 'track' 47.562505 2.170297
' O PGD ' 'nan' 'track' 47.562472 2.17038
' O PGD ' 66.3 'track' 47.562433 2.170453
' O PGD ' 66.3 'track' 47.562395 2.170525
' O PGD ' 66.3 'track' 47.562356 2.170599
' O PGD ' 66.3 'track' 47.562317 2.170672
' O PGD ' 66.3 'track' 47.562278 2.170744
' O PGD ' 66.3 'track' 47.56224 2.170817
' O PGD ' 66.3 'track' 47.562201 2.17089
' O PGD ' 66.3 'track' 47.562163 2.170962
' O PGD ' 66.3 'track' 47.562125 2.171034
' O PGD ' 'nan' 'track' 47.562085 2.171109
' O PGD ' 66.2 'track' 47.562046 2.171182
' O PGD ' 66.3 'track' 47.562008 2.171255

It must look like:

'Flags' 'Dist[km]' 'SoT' 'Lat.' 'Long.'
' O PGD ' 66.4 'track' 47.562584 2.170151
' O PGD ' 66.4 'track' 47.562544 2.170225
' O PGD ' 66.4 'track' 47.562505 2.170297
' O PGD ' 66.3 'track' 47.562433 2.170453
' O PGD ' 66.3 'track' 47.562395 2.170525
' O PGD ' 66.3 'track' 47.562356 2.170599
' O PGD ' 66.3 'track' 47.562317 2.170672
' O PGD ' 66.3 'track' 47.562278 2.170744
' O PGD ' 66.3 'track' 47.56224 2.170817
' O PGD ' 66.3 'track' 47.562201 2.17089
' O PGD ' 66.3 'track' 47.562163 2.170962
' O PGD ' 66.3 'track' 47.562125 2.171034
' O PGD ' 66.2 'track' 47.562046 2.171182
' O PGD ' 66.3 'track' 47.562008 2.171255

Thanks in advance.

[Matz Johansson Bergström]

Could you upload the raw data as a .mat file? It would be so much easier to debug.

[Yuliia]

Here it is

Answer 1

[Andrei Bobrov]

data = raw(:,[3,14,17,20,21]);
ll = ~cellfun(@(x)strcmp('nan',x),data(:,2));

out = data(ll,:);

[Yuliia]

Thanks a lot. It works for me. But, I have a question. What does this line do?

ll=~cellfun(@(x)strcmp('nan',x),data(:,2));

I understand it like it searches for indexes of 'nan' and then command

out = data(ll,:);

displays these values. But, I want to filter by all the conditions at the same time. I use this:

ll=~cellfun(@(x)strcmp('nan',x),data(:,2));
ll2=~cellfun(@(x)strcmp('search',x),data(:,3));
ll3=~cellfun(@(x)strcmp(' O  PGD ',x),data(:,1));
ll4=~cellfun(@(x)strcmp('memory',x),data(:,3));
out4=data(ll&ll2&ll3&ll4,:);

As a result, I don't receive needed values, not all rows with 'memory' deleted. How to write correctly to display data filtered by all these conditions?

[Andrei Bobrov]

Hi Yuliia! For your case:

data = raw(:,[3,14,17,20,21]);
ll = ~any([strcmp('nan',data(:,2)),...
             ismember(data(:,[1,3]),{' O  PGD ','search','memory'})],2);
out = data(ll,:);

or

data = raw(:,[3,14,17,20,21]);
d = data(:,1:3);
ll = false(size(d));
l0 = find(cellfun(@ischar,d));
ll(l0) = ismember(d(l0),]),{' O  PGD ','search','nan','memory'});
out = data(~any(ll,2),:);

Answer 2

[Matz Johansson Bergström]

Andrei beat me to it (again), but here is how you could answer using a loop.

keep_cols = [3, 14, 17, 20, 21];
tmp = raw(:, keep_cols); %copy the columns we wish to keep

j = 1;
for i = 1:size(raw, 1)
    if ~strcmp(raw(i, 14), 'nan') %filter out the not-a-number distance entries
        new_raw(j, :) = tmp(i, :); %copy to new entry in new_raw
        j = j+1;
    end
end

new_raw contains the filtered data. Preferably you would use cellfun, because it is vectorized, so it is usually much faster.