Finding valid sequences in a list based on specific conditions in matlab

I have a list of scores represented as a cell array in MATLAB, and I need to extract valid sequences from this list based on certain conditions. The scores can be one of three stages: 'Rain', 'Cloud', or 'Sun'.

The conditions for a valid sequence are as follows:

For a sequence of 'Sun' stages, the sequence must have a length of at least 7.

For a sequence of 'Rain' or 'Cloud' stages, the sequence must have a length of at least 7, and all stages within the sequence must be either 'Rain' or 'Cloud'.

I have tried implementing the following code, but it does not provide the correct results for all cases:

scores = {'Rain' 'Cloud' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Rain'};

sequence_start = 1;
num_scores = numel(scores);
valid_sequences = [];

for j = 1:num_scores
    stage = scores{j};
    
    if strcmp(stage, 'Sun')
        sequence_length = j - sequence_start + 1;
        
        if sequence_length >= 7
            valid_sequences = [valid_sequences; sequence_start, j];
        end
    elseif ismember(stage, {'Cloud', 'Rain', 'Thunder'})
        sequence_length = j;
        
        if sequence_length >= 7 && all(ismember(scores(sequence_start:j), {'Cloud', 'Rain', 'Thunder'}))
            valid_sequences = [valid_sequences; sequence_start, j];
        end

        sequence_start = j + 1;  % Reset sequence_start
    else
        sequence_start = j;  % Reset sequence_start
    end
end

if (num_scores - sequence_start + 1) >= 7
    valid_sequences = [valid_sequences; sequence_start, num_scores];
end

this gives me the results valid_sequences = 3, 9 which is correct (Sun starts at position 3 and ends at position 9).

However, when I put the following test cases in:

test 1:

scores = {'Rain' 'Rain' 'Cloud' 'Cloud' 'Rain' 'Cloud' 'Sun' 'Cloud' 'Rain'};

I get the answer valid_sequences = 9,9

test 2:

scores = {'Sun' 'Rain' 'Cloud' 'Cloud' 'Rain' 'Cloud' 'Rain' 'Cloud' 'Thunder' 'Sun'};

valid_sequences =

7 7

8 8

9 9

Which suggests there seems to be something wrong with how the Rain/Cloud/Thunder gets read in after Sun.

I would be very grateful if you could help!

Answer 1

[Bruno Luong]

See if the results are what you expect (if not please explain why)

getvalidseq({'Rain' 'Cloud' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Rain'})

ans = 1×2

3 9

getvalidseq({'Rain' 'Rain' 'Cloud' 'Cloud' 'Rain' 'Cloud' 'Sun' 'Cloud' 'Rain'})

ans =

  0×2 empty double matrix

getvalidseq({'Sun' 'Rain' 'Cloud' 'Cloud' 'Rain' 'Cloud' 'Rain' 'Cloud' 'Thunder' 'Sun'})

ans = 1×2

2 9

function valid_sequences = getvalidseq(scores)
b = [true; diff(ismember(scores(:), {'Rain' 'Cloud' 'Thunder'})); true];
i = find(b);
n = diff(i);
j = find(n>=7);
valid_sequences = [i(j) i(j+1)-1];
end

[nines]

getvalidseq({'Rain' 'Rain' 'Cloud' 'Cloud' 'Rain' 'Cloud' 'Sun' 'Cloud' 'Rain'})
ans =

0×2 empty double matrix

this should be hopefully be: 1 6; 7 7

because ideally i will then be able to take these values and extract the corresponding timeseries and merge them. Thanks so much for your help!

[Bruno Luong]

I don't understand, because they have respectively lengths 6 and 1, and you say you only keep those with length >= 7. And why discard |8 9]? There is no coherence in the expected results. The best I can do is:

getvalidseq({'Rain' 'Cloud' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Rain'})

ans = 3×2

1 2 3 9 10 10

getvalidseq({'Rain' 'Rain' 'Cloud' 'Cloud' 'Rain' 'Cloud' 'Sun' 'Cloud' 'Rain'})

ans = 3×2

1 6 7 7 8 9

getvalidseq({'Sun' 'Rain' 'Cloud' 'Cloud' 'Rain' 'Cloud' 'Rain' 'Cloud' 'Thunder' 'Sun'})

ans = 3×2

1 1 2 9 10 10

function valid_sequences = getvalidseq(scores)
b = [true; diff(ismember(scores(:), {'Rain' 'Cloud' 'Thunder'})); true];
i = find(b);
n = diff(i);
%j = find(n>=7);
j = find(n>=0);
valid_sequences = [i(j) i(j+1)-1];
end

[nines]

I am so sorry, that was a typo.

this should be hopefully be: 1 6; 8 9. I would then combine 1-6 and 8-9 and have a total of 8 epochs.

[Bruno Luong]

OK this should do it.

getvalidseq({'Rain' 'Cloud' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Sun' 'Rain'})

ans = 1×2

3 9

getvalidseq({'Rain' 'Rain' 'Cloud' 'Cloud' 'Rain' 'Cloud' 'Sun' 'Cloud' 'Rain'})

ans = 2×2

1 6 8 9

getvalidseq({'Sun' 'Rain' 'Cloud' 'Cloud' 'Rain' 'Cloud' 'Rain' 'Cloud' 'Thunder' 'Sun'})

ans = 1×2

2 9

function valid_sequences = getvalidseq(scores)
b = [true; diff(ismember(scores(:), {'Rain' 'Cloud' 'Thunder'})); true];
i = find(b);
n = diff(i);
nn = size(n,1);
j1 = (1:2:nn)';
j2 = (2:2:nn)';
jj = {j1, j2};
jj = jj(cellfun(@(j) sum(n(j)), jj) >= 7);
j = cat(1,jj{:});
valid_sequences = [i(j) i(j+1)-1];
end

PS: I reread your original question and you never state the merge thing. I'm notorious here to dislike when people asking incompete question, or change the request during the thread is developping.

[nines]

Thank you so so much!!!! I am so sorry for the lack of clarity, I will do better going forward. Thanks so incredibly much, this moves my work forward a ton!!